# Chem(o)Info Radar article snapshot > 309 records returned from 3047 current records. Generated: 2026-09-23T18:29:32.669482+00:00 Filters: days=7 Views are bounded by topic, source, days and limit; text search runs in the dashboard page, not on the server. Blog entries are metadata-only; use their source URL for full text. ## How Do We Process AI Math Advances? - Source: Deep into the Forest (Bharath Ramsundar) (feeds) - Date: 2026-09-23T16:02:13+00:00 - Categories: Blog - Source URL: - Dashboard article: - Abstract: not stored for this record. ## mlxmolkit updated - Source: Macs in Chemistry (Macinchem Blog) (feeds) - Date: 2026-09-23T13:52:43+00:00 - Categories: Blog - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Recent advances in AI-driven biotherapeutic discovery using protein generative models - Source: Digital Discovery (journals) - Date: 2026-09-23T08:40:39Z - Authors: Haelyn Kim, Hakyung Lee, Jinung Song, Juyong Lee - Journal: Digital Discovery - DOI: 10.1039/d6dd00222f - Keywords: generative models - Source URL: - Dashboard article: Abstract: Protein generative models design therapeutic backbones and sequences that are then filtered by cross-validation metrics. We review these methods, their therapeutic applications, and how far these metrics actually predict experimental success. ## Deep learning of tautomer stability from crystallographic proton positions - Source: Chemical Science (journals) - Date: 2026-09-23T07:59:46Z - Authors: Xiaolin Pan, Chao Han, Fengyang Han, Yingkai Zhang - Journal: Chemical Science - DOI: 10.1039/d6sc03714c - Keywords: graph neural network - Source URL: - Dashboard article: Abstract: By mining crystallographic proton positions, we trained a graph neural network to predict stable tautomeric states directly from 2D molecular topology, enabling rapid tautomer assignment for high-throughput molecular discovery. ## Decoding Dopant-Induced Electronic Modulation in Graphene via Region-Resolved Machine Learning of XANES - Source: Digital Discovery (journals) - Date: 2026-09-23T02:43:36Z - Authors: Yinan Wang, Arpita Varadwaj, Teruyasu Mizoguchi, Masato Kotsugi - Journal: Digital Discovery - DOI: 10.1039/d6dd00261g - Keywords: density functional theory, DFT - Source URL: - Dashboard article: Abstract: Revealing how heteroatom doping alters the local electronic structure of graphene is crucial for understanding and controlling its functional properties. In this study, we combine density functional theory (DFT) and... ## A Catalyst Discovery Workflow Coupling MLIP with MOBO for Nitrite-to-Hydroxylamine Electroreduction - Source: ChemRxiv (preprints) - Date: 2026-09-23T00:00:00Z - Authors: Zhiyi Yu, Huaiwen Zhang, Weiming Zhao - DOI: 10.26434/chemrxiv.15009355/v1 - External ID: 10.26434/chemrxiv.15009355/v1 - Keywords: MLIP - Source URL: - Dashboard article: Abstract: The selectivity of electrocatalytic nitrite reduction to hydroxylamine is governed by multiple coupled steps within the multi-electron nitrogen–oxygen reduction network, which cannot be represented simultaneously by a volcano relation based on a single adsorption descriptor. To address this limitation, a catalyst discovery workflow coupling machine-learned interatomic potentials (MLIPs) with multi-objective Bayesian optimization (MOBO) is developed. A small set of physical descriptors maps discrete oxide–dopant candidates onto a low-dimensional continuous representation, while four thermodynamic objectives describe the competing steps in NH2OH formation. The MLIP provides unified, low-cost structural and energy evaluation for oxide candidates. A Gaussian process surrogate model together with expected hypervolume improvement sequentially selects candidates from the discrete pool and updates the Pareto front within a fixed evaluation budget. The workflow integrates chemical objectives, energy evaluation, and sequential decision-making for catalyst screening in oxide systems. ## A general deep learning method based on substrate information for genome-scale mining of natural product biosynthetic enzymes - Source: Journal of Cheminformatics (journals) - Date: 2026-09-23T00:00:00+00:00 - Categories: Cheminformatics - Authors: Yangyang Li, Yuntong Feng, Zhiqiang Wei, Xiangzhao Mao, Hong Jiang, Xiancong Hou, Dawei Yang, Wenzheng Han, Hao Liu - Journal: Journal of Cheminformatics - DOI: 10.1186/s13321-026-01307-1 - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Anyone Can Dock: an online molecular docking tool for everyone - Source: Journal of Cheminformatics (journals) - Date: 2026-09-23T00:00:00+00:00 - Categories: Docking & Screening - Authors: Kowit Hengphasatporn, Thanthapatra Bunchuay, Lian Duan, Yasuteru Shigeta - Journal: Journal of Cheminformatics - DOI: 10.1186/s13321-026-01306-2 - Keywords: molecular docking - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Assessing the state of the art in ADMET prediction: Lessons from the OpenADMET-ExpansionRx community blind challenge - Source: ChemRxiv (preprints) - Date: 2026-09-23T00:00:00Z - Categories: ADMET & Safety - Authors: Maria A. Castellanos, Hugo MacDermott-Opeskin, Jon Ainsley, Jonathan A. Swain, Sean Colby, Devany West, Georgia Channing, Andrew Good, Elyse Bourque, Lakshminarayana Vogeti, Renato Skerlj, Tiansheng Wang, Mark Ledeboer, David Alencar Araripe, Rafal Bachorz, Amitesh Badkul, Maximilian Beckers, Jérémy Besnard, Olivier Beyens, Maicol Bissaro, Davide Boldini, Laetitia Breuil, Jackson W Burns, Lianjin Cai, Ruel Cedeno, Joanna Ceklarz, Xin Chen, Alan C. Cheng, Vishnu Vardhan Chundu, Vladimir Chupakhin, Thomas Coudrat, Simon Crouzet, Daniel Crusius, Wim Dehaen, Valeriia Fil, Robert Fraczkiewicz, Jozef Fülöp, Enrico Gandini, David Giganti, Anthony Gitter, Alec Glisman, Pawel Gniewek, Ryan Greenhalgh, Malte Grieswelle, Oleg Gromov, Sudhir Gupta, Leonard F. Haasbroek, Calum Hand, Xibing He, Dimitrios Iliadis, Abdul Haleem Mohamed Iliiyas, Eachan Johnson, Kalen Josifovski, Shivananda Kandagalla, Lukas Kerti, Oliver Koch, Vaishnavi Kodamala, Stephanie Labouille, Shaolong Lin, Dmitriy Makarov, Tatiana Malygina, James L. McDonagh, Janosch Menke, Linqing Mo, Son Ngo, Ashok Palaniappan, Ferruccio Palazzesi, Csaba Petre, Long-Hung Pham, Will R Pitt, Gopi Krishna Ponnamudi, Suryavedha Pradhan, Eleonora Proia, John Proudfoot, Fabian Rauscher, Jacob Remington, Nicholas Rosa, Tamilazhagan S, Piyush Sawner, Lars L Schaaf, Femi Segun, Dolly Sharma, Harshit Singh, Rajeev Kumar Singh, Ramesh Sistla, Ryan Snyder, Thibaud Southiratn, Satya Pratik Srivastava, Chris Taylor, Igor Tetko, Elisha Gabrielle V. Tiu, Srajal Tiwari, Hunzallah Usmani, Jason Y. Wang, Peng Wang, Shihang Wang, Azmine Toushik Wasi, Brian Wylie, Lei Xie, Yuna Yan, Massa Zahdeh, Shuo Zhang, Donghai Zhao, Mingning (Nancy) Zhu, Martin Šícho, W. Patrick Walters - DOI: 10.26434/chemrxiv.15009342/v1 - External ID: 10.26434/chemrxiv.15009342/v1 - Keywords: ADMET prediction, lead optimization, ADMET - Source URL: - Dashboard article: Abstract: Computational blind challenges are essential for advancing drug discovery, especially in research areas with limited public data, such as absorption, distribution, metabolism, excretion, toxicity (ADMET) prediction. In this paper, we present the results of our latest community blind challenge, organized in collaboration with Expansion Therapeutics (ExpansionRx). Participants were tasked with predicting nine distinct ADMET endpoints routinely evaluated during lead optimization of compounds in drug discovery programs. Following unprecedented engagement from the global community, this challenge provided valuable insights into the state of the art in machine learning for ADMET modeling. Evaluating submissions against a realistic, challenging dataset from active ExpansionRx drug discovery programs revealed that message-passing neural networks and chemical foundation models dominated among the top performers. Furthermore, the results highlighted the power of proprietary industry datasets and a critical need to optimize predictive workflows for different endpoints independently via multitasking and task-affinity grouping. We also demonstrate how hosting our open-source blind challenge infrastructure on a flexible platform like Hugging Face was pivotal for driving community engagement and visibility. Ultimately, blind challenges supported by robust infrastructure, widespread dissemination, and rigorous post-challenge analysis are fundamental to maximizing the utility of public data releases. We anticipate that these community benchmarks will continue to drive the advancement of ADMET models, enabling their deployment in both early- and late-stage drug discovery. ## CAPRICHO: Interpretable Quality Flagging and Flexible ChEMBL Bioactivity Curation for QSAR Modeling - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-23T00:00:00+00:00 - Categories: Cheminformatics, Property Prediction, ADMET & Safety - Authors: David Alencar Araripe, Srijit Seal, Olivier J. M. Béquignon, Gerard J. P. van Westen - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c00802 - Keywords: QSAR, ChEMBL, CYP inhibition, Bioactivity, Caco 2 - Source URL: - Dashboard article: - Code: Abstract: Preparing ChEMBL bioactivity data for quantitative structure–activity relationship (QSAR) modeling requires navigating curation decisions that often lack documentation and are hard to reproduce. We present CAPRICHO (ChEMBL Aggregation Package with Robust Inspection and Curation Handling Options), a Python command-line tool addressing two gaps in current workflows: transparency and flexibility. Unlike approaches that remove problematic data, CAPRICHO flags quality issues while preserving all data for inspection, letting users assess how each curation decision affects cross-assay comparability. Aggregation is configured via customizable grouping columns, supporting both traditional QSAR and emerging assay-aware modeling paradigms, with all parameters recorded in recipe files for reproducibility. Three case studies demonstrate the tool: (i) programmatic support of ChEMBL max-curation standards, (ii) CYP inhibition multitask curation, and (iii) Caco-2 permeability with unit standardization. Together, these show that preserving rather than removing flagged data lets practitioners identify which quality issues most affect comparability and balance data quality with availability for QSAR modeling. CAPRICHO is freely available at https://github.com/CDDLeiden/Capricho ## Cholesterol Hot Spot Automated Mapping Protocol (CHAMP): Identifying Cholesterol-Binding Hot Spots in Membrane Proteins - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-23T00:00:00+00:00 - Categories: Cheminformatics - Authors: David Sotillo-Núñez, Matteo Nardi Cesarini, Gian Marco Elisi, Mattia Bernetti, Giovanni Bottegoni - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c01419 - Source URL: - Dashboard article: Abstract: Biological membranes are dynamic and heterogeneous environments that actively influence protein structure and function. Cholesterol, in particular, has been widely reported to modulate membrane proteins through binding to specific interaction sites. However, the identification of these sites remains challenging due to the intrinsic complexity and dynamics of lipid bilayers, especially when multiple systems and membrane environments are considered. Here, we present CHAMP (Cholesterol hot spot automated mapping protocol), an automated Python-based workflow that combines coarse-grained molecular dynamics simulations with a dual analysis based on contact persistence and spatial density to identify cholesterol-binding hot spots at protein–membrane interfaces. The protocol requires limited user intervention and is designed to enable consistent and reproducible analyses across different targets. We validated CHAMP on the serotonin transporter and the cannabinoid receptor 1 (CB1), where it recovers known cholesterol-binding regions reported in experimental and computational studies. In the CB1 system, the method also identifies a potential additional interaction site. To assess the generality of the approach, we applied the protocol to the CB1 negative allosteric modulator Org27569, showing that it can capture relevant interaction hot spots beyond sterol molecules. Overall, CHAMP provides a practical framework for mapping interaction hot spots at protein–membrane interfaces and may support large-scale and comparative studies of membrane proteins. Limitations of the current implementation and possible extensions are discussed. ## Computationally Accelerated Discovery of Efficient Dual Component White Light Emitters - Source: ChemRxiv (preprints) - Date: 2026-09-23T00:00:00Z - Authors: T.H. Jones, R.S. Horne, D.J. Graham, B.A. Blight, Barry Blight - DOI: 10.26434/chemrxiv.15009366/v1 - External ID: 10.26434/chemrxiv.15009366/v1 - Keywords: computational screening, DFT - Source URL: - Dashboard article: Abstract: This work presents a computationally guided approach to formulating white light emission from a series of Hbonded iridium phosphors and a complementary H-bonding fluorophore. We computationally investigated thirty new Ir(III) complexes that all incorporate a previously studied H-bond rich biguanide functionalised ligand. The results of the computational screening, based on TD-DFT peak emission data and established lifetime and quantum efficiency data in the literature, identified seven Ir(III) candidates suitable for synthesis and further investigation. Of the seven Ir(III) complexes, four were chosen based on their synthetic feasibility and their suitability to further study based on the parameters of the screen. Photophysical investigations of these four candidates with our deep blue emitter showed how three of the four complexes could achieve near ideal white light in the appropriate ratio of the complementary H-bonding fluorophore, while all four hostguest-complements show promising white-light results in polymer (PMMA) films. ## Computer Vision for Predicting Electrochemical Reaction Performance from Electroanalytical Data - Source: ChemRxiv (preprints) - Date: 2026-09-23T00:00:00Z - Authors: Xinran Chen, Sven Erik Peters, Lutz Ackermann - DOI: 10.26434/chemrxiv.15009343/v1 - External ID: 10.26434/chemrxiv.15009343/v1 - Source URL: - Dashboard article: Abstract: Electrosynthesis has surfaced as a key approach in modern synthesis, enabling more sustainable chemical transformations with reduced chemical waste. The accurate prediction of electrosynthesis performance is crucial for accelerating the discovery and design of reactions. Despite indisputable advances, the descriptor-based predictions fall short in representing the full reaction regime due to the high‑dimensional parameter space of electrosynthesis. Cyclic voltammograms, which are typically employed for mechanistic analysis, inherently contain rich information about electrochemical processes. Herein, we applied computer vision techniques to collect a cyclic voltammogram image database and trained a neural network directly on these image data. The resulting models achieved reliable performance in predicting electrosynthesis outcomes, demonstrating the potential of image-based deep learning for electrochemical reaction evaluation and data-driven electrosynthesis studies. ## CRAFT: A Chemical Reaction-Aware Transformer for Synthesizable Molecular Optimization via Multimodal Learning - Source: Journal of Medicinal Chemistry (journals) - Date: 2026-09-23T00:00:00+00:00 - Categories: Docking & Screening, Design de novo - Authors: Runfu Yu, Tongmao Ma, Zian Song, Qiyao Yin, Shuang Wang - Journal: Journal of Medicinal Chemistry - DOI: 10.1021/acs.jmedchem.6c00920 - Keywords: de novo drug design, molecular docking, drug design, as generative, generative models, Transformer, Molecular Optimization, molecular diversity - Source URL: - Dashboard article: Abstract: Ensuring synthetic accessibility remains a major challenge in de novo drug design, as generative models can produce high-scoring molecules that are difficult to synthesize. To address this issue, we propose CRAFT, a reaction-aware molecular optimization framework that integrates Monte Carlo Tree Search (MCTS) with a multimodal strategy network. Unlike sequence-only baselines, CRAFT explicitly encodes two-dimensional molecular topology and atom-level reaction roles, enabling chemically feasible transformations during multi-step optimization. Comprehensive evaluations on DRD2, AKT1, and CXCR4 demonstrate that CRAFT consistently outperforms state-of-the-art methods, generating more high-scoring candidates while improving reaction-constrained synthetic accessibility and molecular diversity. CRAFT also rediscovers scaffolds homologous to known active ligands and identifies novel chemical entities beyond the training distribution, as supported by molecular docking simulations. Moreover, the model provides transparent forward-synthesis pathways for generated candidates, helping bridge in silico molecular design and practical synthesis. ## CrossMol: Cross-Modal Mask-Predict Pre-training For 3D Molecular Data - Source: Bioinformatics (journals) - Date: 2026-09-23T00:00:00+00:00 - Categories: Cheminformatics - Authors: Kangjie Zheng, Junwei Yang, Siyu Long, Wei Ju, Wei-Ying Ma, Hao Zhou, Ming Zhang - Journal: Bioinformatics - DOI: 10.1093/bioinformatics/btag702 - Keywords: SMILES - Source URL: - Dashboard article: - Code: Abstract: Motivation Self-supervised pre-training models for molecular data have demonstrated notable results across many downstream tasks. The inherent multimodal properties of molecules have also motivated efforts to capture information from different modalities. However, current multimodal molecular pre-training models usually treat these modalities as equal and independent, despite differences in their information content. Three-dimensional (3D) molecular structures generally contain finer-grained information than the simplified molecular-input line-entry system (SMILES), which primarily captures higher-level semantic information, such as molecular topology. Results We designed a cross-modal mask-predict pre-training model, CrossMol, to capture semantic associations between modalities with unequal information volumes. The model completes missing 3D structure using higher-level semantic information from another modality, such as SMILES. This allows it to learn cross-modal associations and better understand fine-grained 3D structural information. We also introduce a reweighted distance prediction loss to improve the modelling of short-range structural information. Experiments show that CrossMol achieves large performance gains on multiple downstream molecular tasks, attaining state-of-the-art results. Availability and Implementation The source code and training data are available at https://github.com/zhengkangjie/crossmol. The code is archived at https://doi.org/10.5281/zenodo.19653204. Supplementary Information Supplementary material contains the task-specific hyperparameters. ## Data-Efficient Machine Learning Potentials for Organic Reactions through Active-Learning-Guided Fine-Tuning - Source: ChemRxiv (preprints) - Date: 2026-09-23T00:00:00Z - Authors: Kaipai Ren, Yujing Zhao, Kun Tang, Juntao Wang, Guangran Zhang, Qilei Liu - DOI: 10.26434/chemrxiv.15009333/v1 - External ID: 10.26434/chemrxiv.15009333/v1 - Keywords: Machine Learning Potentials - Source URL: - Dashboard article: Abstract: Transition states (TSs) are central to reaction barriers and mechanisms, yet their accurate localization remains computationally demanding. Reactive machine learning potentials can accelerate TS optimization by rapidly predicting potential energy surfaces, but achieving chemical transferability requires extensive sampling and quantum chemical (QC) labeling of reactive conformations. Here, we introduce an active-learning-guided fine-tuning strategy that couples pretrained chemical knowledge for broad generalization with adaptive labeling of informative reactive conformations for data efficiency. The resulting MACEPre-DeePEST-OS enables efficient TS optimization and barrier predictions across diverse organic reactions while requiring QC labeling for only ~2% of approximately 19 million candidate conformations. Across three in-distribution benchmarks, MACEPre-DeePEST-OS achieves an average root mean square deviation of 0.080 Å for TS geometries and a mean absolute error (MAE) of 0.938 kcal/mol for barriers. In cross-dataset evaluation on an external OrgReact subset, fine-tuning reduces the barrier MAE by 58% relative to the pretrained model. Out-of-distribution testing on a Ugi reaction network containing structures with more than 70 atoms yields a barrier MAE of 1.09 kcal/mol, indicating transfer to molecular sizes and reaction complexity beyond the fine-tuning data. These results establish a data-efficient route toward transferable reactive potentials for quantitative analysis of reaction mechanisms and competing pathways. ## Differentiable mass transfer modeling enables wet lab validated inverse analysis of liquid chromatography - Source: ChemRxiv (preprints) - Date: 2026-09-23T00:00:00Z - Authors: Baochen Li, Zihan Zhou, Yanqing Yu, Sen Lin, Peng Wang, Tianshu Yu, Xiaoxue Wang - DOI: 10.26434/chemrxiv.15009352/v1 - External ID: 10.26434/chemrxiv.15009352/v1 - Source URL: - Dashboard article: Abstract: Liquid chromatography (LC) is essential to organic synthesis and emerging selfdriving laboratories (SDLs), yet LC method development remains cost-intensive and dependent on expert intuition. Mechanistic chromatographic models offer a route to transferable and data-efficient method development, but their practical use is limited by the computational cost of inverse modeling and by the ill-posedness of parameter estimation from experimental chromatograms. Here we report a wet-lab-validated differentiable physics framework for LC inverse modeling. The framework integrates mechanistic column modeling with physics-informed machine learning (PIML) to enable efficient forward simulation and gradient-based inverse estimation of molecular transport and adsorption parameters from gradient-elution chromatograms. The primary PINN surrogate reproduces the mechanistic reference with a root-mean-square error of 10−3 in dimensionless concentration. To address the ill-posed nature of the inverse problem, we introduce an observability-guided experimental-design strategy that quantifies parameter identifiability and incorporates this information into the inverse objective to promote convergence toward the ground-truth parameters. Across 300 dry-lab simulation parameter combinations, the proposed strategy recovered the ground-truth parameters with relative errors ≤ 10% in 88.3% of cases using three gradient programs, compared with only 6.7% using a single gradient program. We further validate the framework against wet-lab LC 1 measurements, achieving an absolute retention-time error ≤ 10 s for 89.3% of samples and establishing a wet-lab database comprising 731 sets of transport and adsorption parameters extracted from 2,553 gradient-elution LC chromatograms. This work provides an observability-aware differentiable modeling foundation for autonomous LC method development and closed-loop analytical decision-making in AI-driven laboratories. ## Docking, Molecular Dynamics, QSAR, and Machine Learning in Alzheimer's Small-Molecule Drug Discovery: Current Utility, Key Limitations, and Reproducibility Challenges - Source: ChemRxiv (preprints) - Date: 2026-09-23T00:00:00Z - Categories: Property Prediction, Docking & Screening, ADMET & Safety - Authors: Poulami Saha, Anuja Chouhan - DOI: 10.26434/chemrxiv.15009354/v1 - External ID: 10.26434/chemrxiv.15009354/v1 - Keywords: QSAR, molecular docking, property prediction, Molecular Dynamics, MD simulations, ADMET - Source URL: - Dashboard article: Abstract: Alzheimer’s disease (AD) remains one of the most challenging therapeutic areas for small-molecule drug discovery because of its multifactorial pathology, complex target biology, and demanding central nervous system drug-development requirements. In this context, computational methods have become increasingly important for compound prioritization, hypothesis generation, and structure–activity interpretation. Among these, molecular docking, molecular dynamics (MD) simulations, quantitative structure–activity relationship (QSAR) modeling, and machine learning (ML) are widely used in AD-oriented medicinal chemistry, particularly in studies involving targets such as β-secretase 1 (BACE1) and tau. This review critically examines the practical roles of these four approaches in AD small-molecule discovery. Rather than presenting them as universally predictive solutions, the review evaluates their principal strengths, major limitations, and current challenges related to validation, interpretability, reproducibility, and experimental translation. Molecular docking remains useful for structure-based hypothesis generation and binding-pose analysis, especially for structurally characterized targets, but is frequently overinterpreted when numerical scores are treated as direct evidence of biological activity. MD simulations provide a complementary framework for examining protein–ligand stability, conformational flexibility, and interaction persistence, although simulation outcomes remain highly dependent on methodological choices and should not be regarded as definitive validation of binding or efficacy. QSAR continues to support ligand-based activity modeling and analogue prioritization, but its utility depends strongly on dataset quality, applicability domain, and validation strategy. ML offers greater flexibility for activity and property prediction, including multi-parameter prioritization and ADMET-related modeling, yet remains sensitive to dataset bias, overfitting, limited interpretability, and poor reproducibility. Across all four approaches, the review emphasizes that methodological rigor is at least as important as methodological sophistication. Particular attention is given to common problems including inadequate validation, overinterpretation of computational outputs, limited reproducibility, and weak integration with experimental medicinal chemistry. Overall, molecular docking, MD simulations, QSAR, and ML are best viewed as complementary support tools whose value depends on careful application, transparent reporting, and close integration with experimental evidence. A more critical, reproducible, and biologically informed use of these methods is likely to provide a stronger foundation for future AD small-molecule discovery. Keywords ## Embedded Morgan Fingerprints for more efficient molecular property predictions with machine learning - Source: ChemRxiv (preprints) - Date: 2026-09-23T00:00:00Z - Categories: Cheminformatics - Authors: Emilio Nuñez-Andrade, Isaac Vidal-Daza, Rafael Gomez-Bombarelli, James W. Ryan, Francisco J. Martin-Martinez - DOI: 10.26434/chemrxiv-2025-6hfp8/v2 - External ID: 10.26434/chemrxiv-2025-6hfp8/v2 - Keywords: molecular fingerprints, Morgan Fingerprint, molecular representations, molecular property - Source URL: - Dashboard article: - Code: Abstract: High-dimensional molecular fingerprints have become a quiet bottleneck for chemical machine learning, where bit collisions in large, chemically diverse datasets and escalating computational costs constrain model discovery. This work introduces the embedded Morgan Fingerprint (eMFP), a post-processing scheme that compresses standard Morgan Fingerprints into continuously valued, lower-dimensional representations governed by a compression factor q. Applied to three molecular datasets of increasing size (RedDB, Non-Fullerene Acceptors and QM9), five regression models and four encoding configurations under scaffold-aware nested cross-validation, eMFP maintains the predictive power of full-length fingerprints for tree-based methods at moderate compression (q≤32), while exhibiting more model- and setting-dependent behaviour elsewhere. Its continuous outputs pair favourably with Fourier Feature Neural Networks, where they interact more naturally with sinusoidal feature mappings than the original binary fingerprints, and consistently shorten training times per trial. Together, these results position eMFP as a simple, drop-in route to more scalable molecular representations that retain accuracy while easing the computational burden of model development. ## Enantioselective C–H Activations by Metallaelectro-Catalysis - Source: Accounts of Chemical Research (journals) - Date: 2026-09-23T00:00:00+00:00 - Authors: Sven Erik Peters, Sven Trienes, Yanjun Li, Lutz Ackermann - Journal: Accounts of Chemical Research - DOI: 10.1021/acs.accounts.6c00509 - Source URL: - Dashboard article: Abstract: Conspectus Given the increasing awareness of the economic and environmental aspects of chemical syntheses, transformative catalysis platforms have emerged, with molecular electrocatalysis residing at the forefront. It enables sustainable oxidative transformations with high levels of energy efficiency and resource economy as well as scalability and inherent process safety. Hence, organic electrocatalysis has gained considerable momentum. Further, in terms of sustainable syntheses, the catalyzed activation of otherwise inert C–H bonds was identified as a particularly powerful strategy for the rapid assembly of chiral scaffolds, thereby preventing stoichiometric chemical waste. Despite indisputable advances, these strategies largely relied, until recently, on stoichiometric quantities of often expensive or toxic chemical oxidants. To this end, metallaelectro-catalysis has emerged as an enabling technology for C–H activations, using the traceless transfer of protons and electrons in lieu of chemical oxidants, while generating molecular hydrogen as the sole by-product via the synchronized hydrogen evolution reaction (HER). Although distinct electrooxidative C–H activations were realized during the past decade, enantioselective electrochemical catalysis has thus far proven elusive. This Account summarizes our key advances until August 2026 in metallaelectro-catalyzed enantioselective C–H activation, streamlining the assembly of chiral molecules featuring various stereogenic elements. Thus, palladaelectro-catalyzed C–H alkenylation set the stage for oxidative C–H/C–H couplings toward atropochiral biaryls, leveraging chiral transient directing groups (TDGs). Subsequent research efforts have allowed us to broaden this platform. Since the pronounced complexity of the electrochemical space in these systems can complicate discovery and optimization, the implementation of a machine learning (ML) workflow was targeted. This was harnessed for the accelerated optimization of an enantioselective palladaelectro-catalyzed annulation reaction. Moreover, the versatility of rhodium catalysis in enantioselective electrochemical C–H activations was mirrored by the e construction of chiral spiropyrazolones and phthalides. Tailored ruthenium catalysts allowed for the selective buildup of atropostable indoles as well as chiral spirocyclic compounds. Taking into consideration the environmental implications and cost-efficacy of transition metal-catalyzed transformations, we further explored the utility of earth-abundant transition metals. On this note, enantioselective cobaltaelectro-catalysis enabled a broad range of highly efficient and enantioselective transformations, such as spirocyclizations, annulations with alkynes, allenes, and alkenes toward axial-, C-, and P-centered chirality, as well as decagram-scale synthesis realized in flow. The data-driven exploration of newly designed chiral ligands facilitated accelerated discovery in cobaltaelectro-catalyzed C–H activation. Moreover, cupraelectro-catalysis was employed for crafting planar chirality in ferrocene derivatives through versatile C–Het bond formations. Finally, we achieved an enantioselective desymmetrization through electrooxidative nickel catalysis inspired by ML, selectively constructing multiple contiguous stereocenters. Interestingly, comparative analysis revealed distinct reactivity for cobalt and nickel catalysts, whose mechanistic origins were elucidated. ## Energy Landscape Sampling Reveals Ligand-Dependent Structural Dynamics of the Protease-Activated Receptor 1 - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-23T00:00:00+00:00 - Categories: Cheminformatics - Authors: Inken Kaja Schwerin, Shihai Jiang, Claudia Stäubert, Berend Isermann, Georg Künze - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c01315 - Source URL: - Dashboard article: Abstract: Protease-activated receptor 1 (PAR1) is a class A G protein-coupled receptor activated by proteolytic cleavage and exposure of a tethered agonist. While existing structures have revealed inactive and active conformations, the ligand-dependent conformational energy landscape governing PAR1 activation remains poorly understood. Here, we employed enhanced-sampling molecular dynamics simulations using the string method with swarms of trajectories to quantify how extracellular ligands and G-protein binding modulate PAR1 activation. Free-energy landscapes were computed for five receptor states (apo, tethered agonist, thrombin receptor-activating peptide (TRAP), vorapaxar (VPX), and VPX + Na+) using five transmembrane (TM) helix distance metrics as collective variables. In the apo state, PAR1 samples inactive, intermediate, and active-like conformations, indicating substantial intrinsic receptor flexibility. VPX increases the free energy of active conformations and restricts TM helix 5 and 6 rearrangements, while VPX + Na+ eliminates intermediate basins and stabilizes the inactive state. In contrast, the tethered agonist stabilizes the active state as the global minimum and promotes TM6 displacement of up to ∼1 nm. The simulations reveal a continuous activation pathway linking extracellular ligand engagement to intracellular motifs, including the connector, DRF, intracellular loop 3 (ICL3), and KRK motifs, with intracellular chloride ion coordination contributing to inactive-state stabilization. Gαq binding reduces TM6 flexibility and shifts the free-energy minimum toward more active-like conformations. Alanine mutagenesis and Gq-CASE BRET assays validate predicted roles of key residues in activation and G-protein coupling. Together, these results provide a detailed understanding of ligand- and transducer-dependent PAR1 signaling and inform the design of pathway-selective PAR1 modulators. ## From Closed to Mighty: How GTP-Induced Allosteric Dynamics, Dimerization, and Membrane Binding Transform a Bacterial Dynamin-like Protein - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-23T00:00:00+00:00 - Categories: Cheminformatics - Authors: Wibke Schumann, Bastian F. Bundschuh, Jennifer Loschwitz, Birgit Strodel - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c01952 - Source URL: - Dashboard article: Abstract: Dynamins, crucial GTPases involved in membrane remodeling, play key roles in processes like endocytosis and membrane fission. While their importance is well-established, the precise dynamics of their conformational changes remain unclear. This study investigates the bacterial dynamin-like protein (BDLP) from Nostoc punctiforme, which undergoes a dramatic closed-to-open transition upon GTP binding. Using standard and enhanced all-atom molecular dynamics simulations, we map this transition, revealing a complex energetic landscape. The initial barrier arises from disruption of salt bridges formed by the GTPase domain with the trunk and paddle, setting off the main hinge motion. Subsequent conformational changes involve further hinge motions and lateral shearing to minimize electrostatic forces. Notably, GTP binding accelerates the transition by 5 orders of magnitude through an allosteric mechanism. Our multiscale simulations of 37.1 μs accumulated simulation time show that BDLP dimerization and membrane binding are necessary to initiate the stabilization of the open conformation. These findings provide crucial insights into BDLP’s membrane remodeling mechanism, advancing our understanding of dynamin dynamics. ## Machine learning-guided drug repurposing for targeting PDK1 in breast cancer treatment - Source: Scientific Reports (journals) - Date: 2026-09-23T00:00:00+00:00 - Authors: Srinivas Ganjipete, Bandral Sunil Kumar, Basavana Gowda Hosur Dinesh, Sameera Hammigi Ramesh, Damodar Nayak Ammunje, Selvaraj Kunjiappan, Parasuraman Pavadai - Journal: Scientific Reports - DOI: 10.1038/s41598-026-72622-8 - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Microchemical Strategies and AI-Enabled Biosensing for Non-Invasive Liquid Biopsy in Precision Oncology - Source: ChemRxiv (preprints) - Date: 2026-09-23T00:00:00Z - Authors: Samuel Long, Abigail Long - DOI: 10.26434/chemrxiv.15009373/v1 - External ID: 10.26434/chemrxiv.15009373/v1 - Keywords: chemometric, Convolutional Neural, CNN - Source URL: - Dashboard article: Abstract: Traditional blood-based liquid biopsy platforms have established crucial paradigms in precision oncology by enabling the longitudinal tracking of circulating tumor DNA (ctDNA) and circulating tumor cells (CTCs). However, blood sampling remains minimally invasive, requires dedicated clinical infrastructure, and is restricted by physiological boundaries such as the blood-brain barrier (BBB). Non-Invasive Liquid Biopsy (NILB) addresses these limitations by utilizing painlessly collected alternative biological matrices, including saliva, urine, breath, stool, sweat, sputum, tears, and cervical secretions. From a microchemical and analytical perspective, these biofluids present extreme analytical challenges: target biomarkers exist at attomolar-to-picomolar concentrations within dense, highly variable chemical matrices containing high background noise and interference. This review outlines the microchemical, physical, and bio-sensing strategies developed to overcome these pre-analytical and analytical barriers. We evaluate advanced microfluidic architectures, nanoscale deterministic lateral displacement, functionalized surface chemistry, isothermal signal amplification, and label-free optical spectroscopy. Furthermore, we evaluate how Artificial Intelligence (AI) and chemometric algorithms—such as Support Vector Machines (SVM), Convolutional Neural Networks (CNN), and multivariate spectral extraction—decode low-abundance, fragmented multi-omic signatures from complex noise. Finally, we propose a standardized analytical framework to bridge point-of-care microchemical devices with routine clinical diagnostic pipelines. ## Nonadiabatic Excited-State Dynamics with Quantum Monte Carlo-Trained Machine Learning: Azomethane as a Stringent Test - Source: Journal of Chemical Theory and Computation (journals) - Date: 2026-09-23T00:00:00+00:00 - Authors: Alfonso Annarelli, Emiel Slootman, Claudia Filippi - Journal: Journal of Chemical Theory and Computation - DOI: 10.1021/acs.jctc.6c01442 - Keywords: force fields - Source URL: - Dashboard article: Abstract: We introduce quantum Monte Carlo (QMC)-trained multistate machine-learned (ML) force fields for nonadiabatic excited-state dynamics, targeting photochemical processes in which the electronic character changes along the reaction path and a consistent correlated description is required. In this framework, variational Monte Carlo (VMC) wave functions combine compact selected configuration-interaction (CIPSI) expansions with a Jastrow factor that explicitly accounts for dynamical correlation, while neural networks convert the stochastic VMC/CIPSI data into smooth potential energy surfaces for large surface-hopping ensembles. We apply this approach to azomethane, a demanding test case involving torsional relaxation through conical-intersection regions and C–N bond dissociation on the hot ground state. Benchmark calculations support the accuracy of the QMC reference data and show robust force convergence across isomerization and dissociation geometries. The QMC-trained dynamics preserves the expected photoisomerization mechanism, strongly reduces the excessive C–N breaking obtained with complete active space self-consistent field, and predicts a small but non-negligible prompt dissociation component after internal conversion, with a time scale consistent with femtosecond-resolved mass-spectrometry experiments. These results establish ML-QMC as a practical route to nonadiabatic photochemical dynamics with accurate wave function reference data. ## Polymer Informatics in the Era of Foundation Models: From Representation to Autonomous Design - Source: ChemRxiv (preprints) - Date: 2026-09-23T00:00:00Z - Authors: Jianxin He, Ying Li - DOI: 10.26434/chemrxiv.15009334/v1 - External ID: 10.26434/chemrxiv.15009334/v1 - Keywords: property prediction, inverse design - Source URL: - Dashboard article: Abstract: Foundation models, pretrained on broad datasets and adapted to multiple tasks, offer a way to reuse chemical knowledge across polymer property prediction and inverse design. Realizing this opportunity requires connecting learned representations to materials whose properties also depend on molecular populations, formulation, processing, and measurement conditions. Here, we review advances in polymer foundation models and the data, representations, and experimental workflows that support their use in materials design. We organize these developments around the polymer material state, a description linking chemical identity to molecular distributions, architecture, morphology, and preparation history relevant to a target property. This organization connects data resources and pretraining strategies to questions about what models learn, how that knowledge transfers, and which experimental choices it can support. We compare learning objectives and adaptation methods, examine representations of polymer chemistry and state, and follow inverse design from candidate generation through synthesis and processing to validation. Pretrained models now support prediction across multiple properties and generation for specified targets, with selected polymer candidates reaching synthesis and thermal characterization. Broader polymer design campaigns show how precursor accessibility, polymerization outcomes, and specimen preparation influence whether a predicted advantage becomes a measured material response. Building on this evidence, we discuss opportunities to combine chemical pretraining with richer sample information and to assess transfer across polymer families, preparation conditions, and experimental campaigns. The resulting synthesis connects model development to a practical objective for polymer informatics: reducing the new experimentation needed to obtain materials with reproducible, targeted performance. ## ProtMutMap: Predicting Protein Binding Free Energy Changes Using Multiple-Mutation Networks - Source: ChemRxiv (preprints) - Date: 2026-09-23T00:00:00Z - Authors: Kairi Furui, Masahito Ohue - DOI: 10.26434/chemrxiv.15009337/v1 - External ID: 10.26434/chemrxiv.15009337/v1 - Keywords: free energy perturbation, Binding Free Energy - Source URL: - Dashboard article: Abstract: Predicting changes in binding free energy ΔΔ G caused by multiple mutations at protein−protein interfaces requires accounting for nonadditive interactions between mutations, which an additive approximation based on single mutations cannot capture, and for the dependence on the computational pathway in stepwise approaches that introduce mutations sequentially. Here, we propose ProtMutMap, a method that integrates information from multiple pathways in a network whose nodes represent target variants and variants containing subsets of their mutations, and whose edges represent free energy perturbation (FEP) calculations for residue substitutions. ProtMutMap simultaneously estimates the binding ΔΔ G of each variant from all edge values using Huber loss. We compared ProtMutMap with the additive approximation and the stepwise method at 30 evaluation points, including intermediate variants, using a multiple-mutation dataset comprising 7 protein complex systems derived from SKEMPI2. ProtMutMap achieved a root mean squared error (RMSE) of 1.13 kcal/mol, lower than that of either comparison method. ProtMutMap also yielded lower prediction errors than Rosetta flex ddG. These comparisons with experimental values demonstrate that integrating multiple pathways through variants containing subsets of the target mutations improves binding ΔΔ G predictions for protein variants with multiple mutations. ## Residue-Conditioned In-Pocket Design for Paralogue Selectivity - Source: ChemRxiv (preprints) - Date: 2026-09-23T00:00:00Z - Authors: Ahmed amine Sadouk - DOI: 10.26434/chemrxiv.15009331/v1 - External ID: 10.26434/chemrxiv.15009331/v1 - Keywords: generative model, receptor - Source URL: - Dashboard article: Abstract: Structure based generative model explore chemical spaces inside the protein active site , typically prioritizing candidates through empirical scoring functions.Few evaluate wthether the results is selective for the intended protein over a close paralogue , and none of them combines that evaluation with residue level conditioning and covalent bounds placement . In this paper we describe a design loop in which each rounds installs one chemical group aimed at one named pocket residue that a pairwise alignment marks as divergent in a chosen anti target , subject while enforcing structural continuity by requiring each successive molecule to retain its predecessor as a substructure. The anti target pocket is built by mutating only the divergent position within the target's own structure , so the two differ at those residues and nowhere else . Across 17 design campaigns the selected scaffold was retained in every final lead . Comparison against 44 reference compounds from the literature that target 12 receptor pair demonstrated preservation of established recognition chemistry in 19 comparisons , against a 12.6% unrelated active decoy rate . generated candidates the structure relationships with established inhibitor , including 78% shared core with imatinib and a 59% with erlotinib while adding alternative chemical strategies . These included residue directed electrophile warhead such as a chloroacetamide targeting ABL1 Cys388, and the generation of a macrocyclic peptidomimetic architecture for KRAS G12C. Together, these results demonstrate an integrated computational framework capable of translating therapeutic objectives and structural evidence into traceable medicinal chemistry design decisions across multiple protein targets. ## StateFlux: Physically Validated Quantum Metabolic State Networks Reveal Limited Incremental Value for Toxicity Prediction - Source: ChemRxiv (preprints) - Date: 2026-09-23T00:00:00Z - Categories: Cheminformatics - Authors: Aris Tsai - DOI: 10.26434/chemrxiv.15009356/v1 - External ID: 10.26434/chemrxiv.15009356/v1 - Keywords: ECFP, Toxicity Prediction, molecular representations, DFT, quantum chemistry - Source URL: - Dashboard article: Abstract: Mechanistic molecular representations are increasingly proposed as a route to improve toxicity prediction by supplementing parent structure with protonation, metabolism, and quantum-chemical reactivity. Whether those physically richer representations provide predictive information that is genuinely orthogonal to molecular structure remains uncertain. Here, StateFlux represents each drug as a network of chemical and predicted metabolic states with node-level GFN2-xTB properties and edge-resolved changes in ionization potential, electron affinity, electrophilicity, solvation, and local reactivity. On a locked DILIrank 2.0 development set (n = 376; 156 DILI-positive and 220 DILI-negative compounds), 10 × 5 scaffold validation gave AUROC/AUPRC values of 0.8249/0.7727 for ECFP, 0.8049/0.7430 for ECFP + FluxCore, and 0.8010/0.7261 for ECFP + StateFluxCore. The FluxCore AUROC decrement relative to ECFP was -0.0199 (95% paired-bootstrap CI, -0.0341 to -0.0072). Strict temporal and scaffold-novel tests likewise showed no consistent advantage, and an independent 5,367-compound genotoxicity benchmark reproduced the result (AUROC 0.7765 for ECFP versus 0.7732 for StateFluxCore). To distinguish limited predictive utility from inaccurate low-cost quantum chemistry, 24 parent-to-metabolite transformations were re-evaluated with 144 ωB97X-D/6-31+G\*/SMD single-point calculations. GFN2-xTB closely tracked DFT changes in electrophilicity (Pearson r = 0.963; MAE = 0.408 eV; 79.2% sign agreement). Thus, transformation-induced quantum reactivity can be physically meaningful without providing reliable incremental information for broad toxicity classification. These results argue that mechanistic fidelity and predictive utility should be validated separately when increasingly elaborate molecular representations are introduced into chemical machine learning. ## Triplet Pruning and Spatial Decomposition in M3GNet: Accuracy and Memory Trade-offs in Scaling Graph Neural Network Potentials - Source: ChemRxiv (preprints) - Date: 2026-09-23T00:00:00Z - Authors: Linh La, Yugam Aggarwal, Truyen Tran, Svetha Venkatesh, Sherif Abdulkader Tawfik - DOI: 10.26434/chemrxiv.15001250/v2 - External ID: 10.26434/chemrxiv.15001250/v2 - Keywords: Neural Network Potentials, Graph Neural Network - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Topology-Stratified Materials Discovery with A Flow-Based Generative Model - Source: arXiv (preprints) - Date: 2026-09-22T15:02:50Z - Categories: Design de novo - Authors: Jingyi Zhou, Oyshee Chowdhury, Noah Oyeniran, Chongze Hu - External ID: 2609.26547v1 - Keywords: Generative Model, generative models - Source URL: - Dashboard article: - PDF: - Code: Abstract: Accurate generation of crystal structures is the foundation to the discovery of high-performance materials for extreme-environment applications, such as aerospace, additive manufacturing, and fusion energy systems. Although generative modeling has emerged as a promising approach for crystal design, its performance remains limited by the complex crystal structures and diverse chemical compositions. In this work, we develop UFO-MGen, a universal flow-based generative model that learns topological features of Wyckoff representations and leverages this information to accurately generate crystals across vast structural and chemical spaces. Compared with state-of-the-art generative models, UFO-MGen achieves the highest crystal generation success rate under a rigorous multi-stability evaluation framework, the highest SUN (stable, unique, novel) rate, and a remarkable extrapolation capability that has not been reported by previous models. Furthermore, a fine-tuning module is implemented to UFO-MGen for property-constrained crystal generation, enabling the inverse materials design toward target properties. The UFO-MGen opens a new avenue for accelerated materials discovery and providing a foundation for universal materials intelligence. ## High-Throughput Virtual Screening - Source: Rowan Newsletter (feeds) - Date: 2026-09-22T13:40:03+00:00 - Categories: Blog - Keywords: Virtual Screening - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Notation matters: cross-representation inconsistency in chemistry language models and its mechanistic origins - Source: Digital Discovery (journals) - Date: 2026-09-22T09:25:04Z - Categories: Property Prediction, LLMs & Agents - Authors: Animesh Mishra - Journal: Digital Discovery - DOI: 10.1039/d6dd00309e - Keywords: LLMs, property prediction, molecular representations, molecular property - Source URL: - Dashboard article: Abstract: Large language models (LLMs) are increasingly applied to molecular property prediction, but whether they encode notation-agnostic molecular representations or exploit surface-level string patterns re- mains untested. We present a dual-phase... ## TCMaster: Confidence-Aware Querying and Workload-Guided Physical Design for Multi-Source Traditional Chinese Medicine Knowledge Graphs - Source: arXiv (preprints) - Date: 2026-09-22T05:38:56Z - Categories: LLMs & Agents - Authors: Zheng Chen, Yuzhu Li, Haoxuan Li, Zhongde Zhang, Lianshun Jin, Peiwu Qin - External ID: 2609.25712v1 - Keywords: LLM - Source URL: - Dashboard article: - PDF: Abstract: Multi-source knowledge graphs (KGs) need query mechanisms that expose reliability and exploit domain structure. This paper presents TCMaster, a property-graph query substrate for confidence-aware traversal and workload-guided physical design over Traditional Chinese Medicine KGs. TCMaster integrates pharmacopoeias, prescriptions, molecular databases, and LLM-extracted micro-semantics into a KG with approximately 221K entities and 723K base edges. It annotates edges with provenance-level confidence, rewrites Cypher queries with confidence predicates, ranks multi-hop paths under PRODUCT, MIN, or weighted-average policies, and uses ontology skew through direction selection, herb-attribute bitmaps, and materialized shortcut edges. On Neo4j, direction selection improves attribute lookup by a factor of 1.47, shortcuts accelerate high-fanout target counting by a factor of 4.42, confidence filtering removes 39.3 percent of low-quality heterogeneous paths, and KG retrieval improves TCMbench QA accuracy by 20.0 percentage points. ## ArticleMiner: Ontology-Guided Knowledge Graph Construction from Scientific Publications - Source: arXiv (preprints) - Date: 2026-09-22T03:01:47Z - Categories: LLMs & Agents - Authors: Md Abrar Jahin, Craig A. Knoblock, Jay Pujara - External ID: 2609.25607v1 - Keywords: LLM - Source URL: - Dashboard article: - PDF: - Code: Abstract: Scientific papers keep much of their quantitative content in tables and supplementary files, where a number means something only through its header, caption, unit, analytical method, and the conventions of its field. Recovering the rows and columns of a table is therefore not the same as recovering the scientific fact it reports. Most semantic table-interpretation methods assume that a clean table is already available and subsequently map its cells or columns to ontology terms, whereas most publication-level extraction systems are designed for a single domain. We study a middle path: a shared process that reads a paper and its supplementary files, gathers evidence from several parsers and a language model, and reconciles that evidence, while a bounded human-authored task module for each task supplies the domain meaning. The module lists the canonical names the graph may use, the surface forms that map to them, a small set of derivation rules and validity constraints, an identity key, and the bindings used to write RDF. It defines what a task is allowed to emit; it does not try to list every convention of a field. We build four such modules (for drug-discovery chemistry, materials science, machine learning, and mineral geochemistry) in the ArticleMiner framework, and evaluate them on 163 papers, including a new geochemistry benchmark with expert-curated ground truth. In comparisons against a same-LLM few-shot baseline, the point estimates favor ArticleMiner on all four tasks, with uncertainty on the two smaller benchmarks. The geochemistry comparison also includes access to supplementary files, so its improvement cannot be attributed to domain guidance alone. ## Accelerating the Design of New Natural Deep Eutectic Solvents with Supervised Variational Autoencoders - Source: ChemRxiv (preprints) - Date: 2026-09-22T00:00:00Z - Authors: Rachid Laref, Mehdi Oubahmane, Najwa Harrati, Nicolas Rolland, Véronique Nardello-Rataj, Adlane Sayede - DOI: 10.26434/chemrxiv.15008532/v2 - External ID: 10.26434/chemrxiv.15008532/v2 - Keywords: generative models, property prediction, variational autoencoder, molecular descriptors - Source URL: - Dashboard article: Abstract: The rational design of natural deep eutectic solvents (NaDESs) is hindered by the vast combinatorial space arising from the possible combinations of hydrogen bond donors and acceptors. This enormous design space limits the systematic experimental exploration of their physicochemical properties. In this work, we propose a supervised variational autoencoder (SVAE) framework to efficiently explore this chemical space by coupling generative modeling with property prediction. The model is trained on a curated dataset of deep eutectic solvents to learn simultaneously the latent structural representations and correlations with target properties, including melting point, density, and viscosity. The SVAE generates virtual NaDES candidates characterized by molecular descriptors, component ratios, and target property values. To associate these virtual candidates with real molecules, a descriptor matching strategy is applied using the COCONUT natural product database. The resulting NaDES candidates are subsequently validated using predictive models trained on experimental DES data. The results show strong agreement between the SVAE-generated target property values and the corresponding predictions of the regression models (R2 = 0.89, RMSE=21.7 K for melting point; R2 = 0.91, RMSE = 0.04 g·cm-3 for density; and R2 = 0.89, RMSE=0.47 cP for ln(viscosity)). This workflow demonstrates how guided generative models can be combined with machine learning models to accelerate the discovery of sustainable solvents. Beyond NaDESs, the proposed strategy provides a general and transferable framework for the rational design of other chemical systems, where combinatorial complexity limits conventional experimental approaches. ## Atomic Environment Aware Lennard-Jones Parameterization for Electrostatic Embedding ML/MM Simulations - Source: ChemRxiv (preprints) - Date: 2026-09-22T00:00:00Z - Authors: João Morado, Kirill Zinovjev, Lester O. Hedges, Julien Michel, Daniel J. Cole - DOI: 10.26434/chemrxiv.15004404/v2 - External ID: 10.26434/chemrxiv.15004404/v2 - Keywords: force field, force fields - Source URL: - Dashboard article: Abstract: Hybrid machine learning/molecular mechanics (ML/MM) methods have emerged as a powerful approach to multiscale modelling. Recent advances have focused on electrostatic embedding schemes, in which atoms in the ML region are polarized by the surrounding MM environment. Consequently, ML/MM electrostatics have been the subject of extensive study, whereas dispersion and close-range repulsive interactions remain comparatively underexplored. In analogy to established QM/MM practice, exchange-repulsion and dispersion interactions are commonly described using Lennard-Jones (LJ) 12-6 potentials with parameters transferred from existing force field libraries. However, such transferable parameters may be suboptimal for ML/MM interfaces, where consistency with the electrostatic embedding model is desirable. Here, we develop methodologies for deriving dynamic LJ parameters based on the exchange-hole dipole moment (XDM) model and the Fedorov polarizability scaling relation. Moreover, we propose a one-shot procedure for fine-tuning LJ parameters to experimental free energies. Focusing on the electrostatic machine learning embedding (EMLE) scheme, we show that these approaches yield EMLE models with competitive accuracy relative to conventional force fields and mechanical embedding schemes, while additionally providing chemically environment-aware LJ parameters. This capability offers a route to improved accuracy in demanding applications such as reactive modelling, where dispersion coefficients may depend strongly on the local chemical environment. ## Boundary-driven phase transformation in layered 3R In2Se3 captured by long range machine learning molecular dynamics - Source: ChemRxiv (preprints) - Date: 2026-09-22T00:00:00Z - Authors: Igor Evangelista, Michaela Cohen, Atul C. Thakur, Anderson Janotti, Chris Benmore, Tingyi Gu, Ganesh Sivaraman - DOI: 10.26434/chemrxiv.15009227/v1 - External ID: 10.26434/chemrxiv.15009227/v1 - Keywords: MACE, molecular dynamics, PBE - Source URL: - Dashboard article: Abstract: Layered In2Se3 switches between closely related crystalline van der Waals states, but its atomistic thermal pathway in rhombohedral 3R geometries remains unresolved because middle-Se motion within each quintuple layer (QL), inter-QL glide, freesurface morphology, and weak interlayer binding evolve together. Here we develop a PBE+D3(BJ)-trained MACERSGA interatomic potential that adds a reciprocal-space gated-attention long-range correction to a short-range MACE backbone. Compared with MACE trained on the same data, MACERSGA reduces test-set energy and force RMSEs by 83.8% and 17.7%, respectively, while preserving the PBE+D3(BJ) exfoliation binding curves of α-and β-like structures. Large-cell molecular dynamics examines 1QL vacuum slabs, 3QL vacuum slabs, and periodic 3QL cells containing one full 3R repeat. Heating and cooling ramps at 5 and 1 K ps−1, supported by fixed-temperature holds, reveal a boundary hierarchy in the finite-rate structural-change window. The 1QL vacuum slab shows structural change first, followed by the 3QL vacuum and periodic 3QL cells. The 3QL vacuum slab also undergoes substantially larger inter-QL glide than the periodic cell. Cooling the α-heated structures directly tests whether the transformed states retain boundary-specific structural memory on the simulated timescale. Upon cooling, both 3QL systems retain transformed local environments near θ ≈ 57◦ at room temperature rather than returning to high-θ α, yet preserve distinct inter-QL registries and out-of-plane morphologies inherited directly from their boundary-specific heating pathways. These results show how nanoscale boundary conditions shape transformation pathways and retained interlayer registry in layered In2Se3, revealing structural memory relevant to nonvolatile crystalline memory and neuromorphic switching concepts. ## Can AI Help Chemists To Solve NMR? - Source: ChemRxiv (preprints) - Date: 2026-09-22T00:00:00Z - Categories: LLMs & Agents - Authors: Fedor Kliuev, Ivan Smirnov, Mikhail A. Losev, Oleg I. Afanasyev, Denis Chusov - DOI: 10.26434/chemrxiv.15006195/v2 - External ID: 10.26434/chemrxiv.15006195/v2 - Keywords: LLMs, Claude, DeepSeek, LLM, GPT - Source URL: - Dashboard article: Abstract: Chemists read 1 H and 13 C NMR daily, and it is tempting to paste the peak lists into a large language model. Given only those two peak lists for 105 molecules balanced by molecular complexity and compound classes, six frontier LLMs (Claude Opus 4.8, DeepSeek V4 Pro, Gemini 3.1 Pro, GPT-5.5, Grok 4.3, Qwen3.7-Max) and four specialized solvers (BLIND, NMRMind, NMRPeak, NMR-Solver) were compared. The best LLM (Gemini 3.1 Pro) solved 24% of structures, the best specialized model (BLIND) solved 69% of them. All models drift toward smaller molecules and drop groups silent in ¹H NMR like halogens, nitriles, diazo. ## Chemical Structures Cleaning Center: An Open Web Service for Automated Chemical Databases Curation - Source: ChemRxiv (preprints) - Date: 2026-09-22T00:00:00Z - Categories: Cheminformatics - Authors: Carlos D. Ramírez-Márquez, José L. Medina-Franco - DOI: 10.26434/chemrxiv.15009294/v1 - External ID: 10.26434/chemrxiv.15009294/v1 - Keywords: chemoinformatics, property prediction - Source URL: - Dashboard article: - Code: Abstract: Chemical database curation is an essential step in chemoinformatics workflows. Omitting this process or using inadequate curation protocols adversely affects structure-dependent computational methods that rely on high-quality data, such as quantitative structure–activity relationship models, physicochemical property prediction, and molecular similarity calculations. Although database curation is compulsory as machine learning, chemoinformatics, and computational chemistry in general rely on data quality, existing public databases curation pipelines are predominantly developed in coding environments like Python and hosted on different servers. While efficient, these tools often present a steep learning curve for researchers without programming expertise or domain specialists seeking to integrate chemoinformatics into their projects. To bridge this gap, herein we present Chemical Structures Cleaning Center, an automated, user-friendly, and open-access web application designed for streamlining chemical database curation. The tool builds upon well-validated and used protocols to curate chemical libraries. To illustrate the platform's utility, we used a small chemical database containing reference compounds to test various pipeline tasks, such as counterion desalting, duplicate detection, small-molecule salt handling, preservation of complex stereochemical integrity, and reduction of metal-organic salts. The curation pipeline provides references to relevant manuscripts highlighting the importance of chemical database curation. The web page is freely available at https://huggingface.co/spaces/Dannmarquez16/Chemical-Structures-Cleaning-Center ## Computer Learning Can Streamline Development of Lipid Nanoparticle Vaccine Delivery Constructs - Source: ChemRxiv (preprints) - Date: 2026-09-22T00:00:00Z - Authors: Eric H. Rosenn, Max Engel, Violet Munarova, Maya Lipshitz, Nata Edell, Dania Halperin, Eryn Fox - DOI: 10.26434/chemrxiv.15009316/v1 - External ID: 10.26434/chemrxiv.15009316/v1 - Source URL: - Dashboard article: Abstract: Lipid nanoparticle (LNP) delivery constructs provide a precise and targeted therapeutic platform for nucleic-acid and subunit vaccines. By shielding fragile antigens and genetic cargo from premature degradation, these constructs help vaccines overcome the physiological barriers that separate an injection site from its intracellular target, increasing specificity, potency, and the precision of controlled release. This capability has broadened the range of immunomodulatory targets in the fight against infectious disease and has opened the door to vaccines that treat tumors and cancer. Increasingly, artificial intelligence (AI) and machine learning (ML) are being applied to predict the LNP formulations that maximize mRNA-vaccine efficacy, and the same computational strategies are now being extended across diverse vaccine modalities. By approximating in vivo pharmacokinetics and pharmacodynamics, in silico methods promote rational, data-driven design that is reshaping the drug-development pipeline. In this Review we summarize current literature on the optimization of LNP delivery constructs for vaccine therapeutics using ML, survey the rapid clinical translation of LNP-mRNA technology since 2021, and discuss the limitations and predictive capabilities of computer-simulated experimental models. ## Discovery of INCB127443: A Potent and Orally Available Inhibitor of Cyclin-Dependent Kinase 2 - Source: Journal of Medicinal Chemistry (journals) - Date: 2026-09-22T00:00:00+00:00 - Categories: Cheminformatics - Authors: Ken Mukai, Brandon R. Smith, Qinda Ye, Leslie B. Epling, Chengtsung Lai, Lu Huo, Sean Bowen, Katherine Drake, Keith Kennedy, Jason Boer, Derek Zimmer, Wenliang Zhang, Michael J. Hansbury, Saswati Chand, Jingwei Li, Guofeng Zhang, Kristine Stump, Yvonne Lo, Margaret Favata, Gengjie Yang, Maryanne Covington, Marc C. Deller, Ricardo Macarron, Jeff Jackson, Patrick Mayes, Sunkyu Kim, Xiaozhao Wang, Joshua R. Hummel, Liangxing Wu, Wenqing Yao - Journal: Journal of Medicinal Chemistry - DOI: 10.1021/acs.jmedchem.6c01506 - Keywords: CDK, pharmacokinetic, Kinase - Source URL: - Dashboard article: Abstract: CDK2 is a critical regulator of cell-cycle progression and is an attractive therapeutic target for cancers characterized by dysregulated cyclin E1 signaling. Despite this biological rationale, CDK2 has historically been challenging to target due to the difficulty of achieving selectivity over related CDK isoforms, particularly CDK1, where off-target inhibition may limit the therapeutic window. Building on our previously reported (4-pyrazolyl)-2-aminopyrimidine series, we pursued a multiparameter optimization strategy aimed at improving selectivity for CDK2 over CDK1 while balancing pharmacokinetic and physicochemical properties. Computationally guided replacement of the pyrazole with an imidazole led to a (4-imidazolyl)-2-aminopyrimidine series and culminated in the discovery of INCB127443 (30), a potent and orally bioavailable CDK2 inhibitor. INCB127443 combined favorable selectivity over CDK1 with a relatively flat pharmacokinetic profile, supporting sustained projected CDK2 coverage while limiting potential CDK1 engagement. In vivo, INCB127443 demonstrated dose-dependent inhibition of Rb phosphorylation and antitumor efficacy in CCNE1-high ovarian cancer xenograft models. ## Enhancing Solubility Predictions of Plastic Additives in Depolymerization Solvents Using Quantum Chemical Conformer Analysis and COSMO-RS - Source: ACS Omega (journals) - Date: 2026-09-22T00:00:00+00:00 - Categories: Cheminformatics, Property Prediction - Authors: Amin Shariatmadar Tehrani, Ruben Goeminne, Pieter Cnudde, Veronique Van Speybroeck, Kathy Elst, Karolien Vanbroekhoven, Elias Feghali, Kevin M. Van Geem, Steven De Meester - Journal: ACS Omega - DOI: 10.1021/acsomega.6c04278 - Keywords: QSPR, RDKit, DFT, Gibbs free energy - Source URL: - Dashboard article: Abstract: The success of recycling plastics by means of depolymerization requires the selective purification of regenerated monomers from complex mixtures. This can be assisted by accurate prediction of the phase behavior of large, flexible, and multifunctional additive molecules, which remains a challenge for current thermodynamic models. In this study, the solubility of 11 representative additives and three PET monomers was measured in two common depolymerization solvents, water and ethylene glycol, covering a wide range from practically insoluble (<1 ppm) to 41.6 wt %. To benchmark predictive capabilities, a quantum-chemistry workflow was investigated, integrating RDKit conformer generation, Gaussian DFT optimizations, and COSMO-RS thermodynamic modeling. Systematic conformer sampling proved critical for bulky, high-molecular-weight additives such as antioxidants. A comparison of three Pople-type basis sets revealed a trade-off between computational cost and accuracy. The 6–311++G(d,p) basis set provided the best balance (RMSE = 1.31 log units), while 3–21+G(d) reduced computational cost at lower accuracy, and 6–311++G(3df,3pd) yielded only minor improvements for select cases at significantly higher cost. Prediction accuracy was highest for more soluble compounds but decreased with increasing molecular weight, conformational flexibility, and Gibbs free energy of fusion. To improve solid–liquid equilibrium predictions, melting points and fusion enthalpies were measured using DSC and compared with QSPR estimates. This combined computational-experimental approach supports the development of energy-efficient purification strategies in chemical recycling of condensation polymers such as PET. ## EVALUATING MOLECULAR DYNAMICS FRAMES FOR BINDING AFFINITY PREDICTION AND VIRTUAL SCREENING - Source: ChemRxiv (preprints) - Date: 2026-09-22T00:00:00Z - Authors: Jakub Poziemski, Pawel Siedlecki - DOI: 10.26434/chemrxiv.15009259/v1 - External ID: 10.26434/chemrxiv.15009259/v1 - Keywords: VIRTUAL SCREENING, MOLECULAR DYNAMICS, BINDING AFFINITY - Source URL: - Dashboard article: Abstract: Motivation: Machine-learning scoring functions are typically trained on static crystallographic structures, although protein-ligand recognition is inherently dynamic. It remains unclear whether MD derived conformations provide transferable information or instead shift the training data away from experimentally observed binding geometries. Results: We generated 20 ns molecular dynamics trajectories for 2,502 protein–ligand complexes and evaluated five scoring function architectures trained on crystallographic structures and MD-derived frames using four different sampling protocols. Across the CASF-2016 and CASF-2013 affinity benchmarks, MD-only training generally reduced predictive performance, whereas augmenting crystallographic structures with MD data improved four of the five architectures. Averaged across models, benchmarks, and sampling protocols, augmentation increased PCC by 0.072 and reduced RMSE by 0.165 relative to MD-only training. A nearest-neighbor analysis showed that the benefits of augmen-tation were not restricted to complexes similar to frames, arguing against a simple similarity-based explanation. The results suggest improvements were connected with exposing models to controlled structural perturbations around experimentally observed binding poses. Overall, MD conformations were most effective as augmentation rather than replacement for crystallographic training data. Availability: MD frames used for model training are available at: (https://zenodo.org/ records/20443988). ## GDEE: A Structure-Based Platform for Gene Discovery and Enzyme Engineering - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-22T00:00:00+00:00 - Categories: Docking & Screening, ADMET & Safety, Targets & Structures - Authors: Caio S. Souza, João P. G. Correia, Isabel Rocha, Diana Lousa, Cláudio M. Soares - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.5c02688 - Keywords: Enzyme - Source URL: - Dashboard article: Abstract: Enzymes are extensively utilized for catalyzing the production of value-added compounds in diverse sectors with significant economic impact. However, challenges such as poor protein expression, low catalytic activity, substrate/co-factor limitations, and toxicity of final products limit the efficiency of these biocatalysts. Protein engineering offers a solution by redesigning enzyme catalytic properties to enhance biosynthetic pathways. In this study, we present an automated platform for gene discovery and enzyme engineering, aimed at overcoming these bottlenecks. The GDEE platform finds and optimizes enzymes responsible for rate-limiting steps in biosynthetic pathways, with objectives ranging from improving catalytic efficiency to enabling novel transformations. The platform operates through four key steps: an initial sequence step that either sources natural enzyme sequences or generates mutant variants, followed by an atomistic protein structure prediction step, a docking step, and a binding energy evaluator, obtaining a set of variant enzyme sequences that are prioritized for experimental validation. Additionally, machine learning-based re-scoring of binding poses to improve binding affinities significantly enhances the accuracy of predicting the effect of mutations on binding, highlighting its potential in metabolic engineering applications. ## Generative AI designs functional thiolation domains for reprogramming non-ribosomal peptide synthetases - Source: Nature Communications (journals) - Date: 2026-09-22T00:00:00+00:00 - Categories: LLMs & Agents - Authors: Emre F. Bülbül, Seounggun Bang, Kevin George, Gabriele Bianchi, Prateek Raj, Seonyong Chung, Vincent Pauline, Ramon Hochstrasser, Hannah A. Minas, Walid A. M. Elgaher, Andreas M. Kany, Anna K. H. Hirsch, Steven Schmitt, Dirk W. Heinz, Olga V. Kalinina, Dietrich Klakow, Kenan A. J. Bozhüyük - Journal: Nature Communications - DOI: 10.1038/s41467-026-77963-6 - Keywords: generative models, molecular dynamics - Source URL: - Dashboard article: Abstract: Large language models and generative protein design promise to accelerate biotechnology, but it remains unclear whether they can engineer dynamic megasynth(et)ases whose activity depends on transient, context-specific domain interfaces. Non-ribosomal peptide synthetases (NRPSs) exemplify this challenge and produce many clinically used therapeutics. Here we integrate pretrained generative models (ESM3, ProteinMPNN and EvoDiff) with design–build–test–learn cycles and data-guided prioritization to generate 76 de novo thiolation (T) domains. We build and test 578 recombinant NRPS variants in vivo spanning minimal, full-length, and hybrid assembly lines. AI-designed T-domains support product formation across architectures, enable catalytically active hybrids at recombined junctions, and increase yields by up to ~3-fold relative to NRPSs carrying the native T-domain. A representative design shows improved soluble expression, refolding, and a 12 °C higher melting temperature, while molecular dynamics simulations indicate preserved global stability but reshaped, state-dependent interdomain contact networks. Together, these results establish generative design as an effective route to context-conditioned engineering and reprogramming of biosynthetic assembly lines. ## Hierarchical Phase Transitions in Bayesian Flow Networks Enable Training-Free Molecular Design - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-22T00:00:00+00:00 - Categories: Docking & Screening, Design de novo - Authors: Liangji Zhou, Yuanqiu Chen, Tongtong Chen, Yijuan Huang, Qihao Pan, Si Li, Chenyu Gou, Meiqiong Fan, Yu Gao - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c01538 - Source URL: - Dashboard article: Abstract: Bayesian flow networks (BFNs) jointly model atomic coordinates and atom types for structure-based drug design, but their generative dynamics remain poorly understood. We show that the two channels resolve on different terms: atom types pass through a sharp commitment transition at tc1, whereas coordinate precision improves smoothly with no critical point of its own, reaching sub-Ångstrom resolution only at a stated threshold tc2. Extreme-value theory predicts the type channel’s posterior half-maximum in closed form, and the predicted 1/β1 scaling holds across seven β1 values (R2 = 0.996) and over 13 points spanning three BFN architectures and two domains, though measured times run 11%–20% below the predicted absolute values. Exploiting that structure, phase-aware iterative refinement aligns three rounds to these boundaries and trains no auxiliary regressor, reward function, or property predictor. On 100 CrossDocked2020 test pockets at K = 200, it reaches a mean Vina Dock of −8.72 kcal/mol. Gradient-guided methods that need such training stay ahead by a small margin: over the 100-pocket intersection, the paired gap to MolJO (−8.98) is 0.25 kcal/mol (95% CI \[−0.13, +0.60\], paired Wilcoxon p = 0.022); equivalence testing at a preregistered ±0.5 kcal/mol margin does not establish equivalence within it. CByG (−9.16) leads by 0.44 kcal/mol on unpaired means. Phase-aware iteration improves the Round-1 to Round-3 mean by 1.34 kcal/mol (paired Wilcoxon p = 4.80 × 10–18, Cohen’s dz = −1.59) and shifts the median by −1.44 kcal/mol, a shift Best-of-K resampling cannot produce. Paired on the pocket, an audited rerun passes every PoseBusters check 4.7 percentage points more often than MolJO. ## Highly Efficient and Water‐Resistant Copper–Iodide Emitters Toward Underwater X‐Ray Imaging - Source: Angewandte Chemie International Edition (journals) - Date: 2026-09-22T00:00:00+00:00 - Authors: Xue Gao, Yonghao Zhu, Zhiyuan Wu, Jieping Zhang, Jieshi Mai, Yongsheng Liu, Lian Chen, Maochun Hong - Journal: Angewandte Chemie International Edition - DOI: 10.1002/anie.7210262 - Keywords: RL - Source URL: - Dashboard article: Abstract: Advanced scintillators with superior performance and environmental robustness are urgently needed for emerging x‐ray imaging applications, especially in challenging environments such as underwater scenarios. Copper(I) iodide cluster complexes are promising candidates for advanced x‐ray imaging applications due to their excellent optical properties and solution‐processable nature. Herein, we report two green‐emitting Cu 2 I 2 cluster‐based complexes achieved through a dual‐ligand strategy. The rigid P,N‐bridging ligand serves to construct the emissive Cu 2 I 2 core and improve exciton utilization, while the meta‐fluorinated ancillary ligand introduces multiple weak interactions, suppressing nonradiative decay and enhancing water stability. Together with the aggregation‐induced emission (AIE) characteristics, two complexes achieve high photoluminescence quantum yields (PLQYs) of 86.4% and 92.5% in the solid state, respectively. The optimal complex shows efficient radioluminescence (RL) with a low detection limit of 73.5 nGy air s −1 . The fabricated flexible screen exhibits stable x‐ray imaging performance with a spatial resolution up to 21.4 lp mm −1 . More importantly, the complex demonstrates exceptional structural integrity and emission stability in air and water. These attributes ultimately allow for high‐performance underwater x‐ray imaging. This work not only provides a ligand‐engineering strategy for the construction of high‐PLQY, water‐resistant copper‐cluster‐based scintillators, but also advances their practical application in emerging x‐ray imaging scenarios. ## Identification of Gene Signatures and Molecular Mechanisms Underlying the Comorbidity of Alzheimer's Disease and Crohn's Disease Using Machine Learning. - Source: Psychiatry investigation (journals) - Date: 2026-09-22T00:00:00Z - Categories: Docking & Screening - Authors: Hong-Wei Liu, Zhao-Xu Yin, Zhi-Nan Ye, Hao Xu - Journal: Psychiatry investigation - DOI: 10.30773/pi.2026.0214 - External ID: fb82e49f4340b840977ef82eb0d47a5c9a1add48 - Keywords: Molecular docking, binding affinity, receptor - Source URL: - Dashboard article: Abstract: OBJECTIVE Alzheimer's disease (AD) and Crohn's disease (CD) both involve inflammation and immune dysregulation, yet the potential molecular mechanisms underlying their comorbidity remain unclear. METHODS We integrated transcriptomic data from AD and CD patients and applied differential expression analysis, weighted gene coexpression network analysis, protein-protein interaction networks, and multiple machine learning approaches to identify key comorbidity genes. Functional enrichment, single-cell sequencing validation, and virtual knockout analyses were used to explore their biological roles. Molecular docking was performed to evaluate the binding affinity of candidate small-molecule drugs to the identified core genes. RESULTS CXCL1 and IGFBP5 were identified as core comorbidity genes. CXCL1 was associated with inflammatory signaling, including cytokine receptor binding, neutrophil migration, and NOD-like receptor signaling. IGFBP5 was linked to growth factor binding, smooth muscle cell proliferation, and extracellular matrix-receptor interactions. Single-cell and virtual knockout analyses indicated that these genes play pivotal roles in inflammation, immune regulation, cell migration, and tissue remodeling, potentially bridging central and peripheral inflammation via the gut-brain axis and IgSF CAM signaling. Candidate drug prediction and molecular docking suggested that small molecules such as Dasatinib, Mifepristone, and Retinoic acid may modulate these pathways. CONCLUSION This study reveals the critical roles of CXCL1 and IGFBP5 in AD-CD comorbidity, providing a theoretical basis for exploring gut-brain axis mechanisms and potential targeted interventions. ## Interactions between Tick-Borne Encephalitis Virus Nonstructural Protein 1 and Blood–Brain Barrier Tight Junction Proteins: Potential Clues to Strain-Specific Neuropathogenicity - Source: ACS Omega (journals) - Date: 2026-09-22T00:00:00+00:00 - Categories: Targets & Structures - Authors: Jarinya Chaopreecha, Jordan Liburd, Shahid Khalid, Songeziwe Ntsimango, Ian M. Jones, Jennifer H. Tomlinson, Niluka Goonawardane - Journal: ACS Omega - DOI: 10.1021/acsomega.6c00564 - Keywords: AlphaFold3, AF3 - Source URL: - Dashboard article: Abstract: Tick-borne encephalitis virus (TBEV) invades the central nervous system (CNS) through strain-specific mechanisms that remain poorly understood. For mosquito-borne orthoflaviviruses such as dengue and yellow fever viruses, the nonstructural protein 1 (NS1) has been shown to disrupt endothelial barrier integrity by targeting tight junction proteins (TJPs), facilitating viral neuroinvasion. However, comparable mechanisms in TBEVs remain largely unexplored. Here, we investigated the potential interaction of NS1 from high (Hypr)- and low (Vs)-pathogenic TBEV strains with different blood–brain barrier (BBB) TJPs, using AlphaFold3 (AF3) multimer modeling and in vitro binding assays. AF3 modeling suggested that NS1 from the highly pathogenic strain may interact with multiple TJPs, including junctional adhesion molecule A (JAM-A), a key component of the paracellular barrier. In contrast, the low-pathogenic Vs showed a more restricted interaction profile. Experimental validation using recombinant NS1 proteins revealed strain-specific binding profiles: Hypr NS1 displayed high-affinity, saturable direct binding to immobilized JAM-A (KD,app = 0.81 nM), whereas Vs NS1 showed no binding. Immunofluorescence assays on human lung epithelial (A549) and BBB (hCMEC/D3) cells further demonstrated colocalization between Hypr NS1 and JAM-A, supporting a potential interaction. No direct binding to ZO-1, a barrier scaffold lacking an extracellular domain, was observed for either strain. This differential interaction profile may be modulated by 22 amino acids localized to the Wing and β-Ladder domains within NS1, which distinguish Hypr from Vs. Notably, despite its limited interactions with the investigated TJPs, the Vs strain is associated with slow-progressive infections that can culminate in chronic neurological disease. This highlights the need for further studies into noncanonical pathways of neuroinvasion. ## Intuitive enzyme design with LLM agents - Source: Nature Computational Science (journals) - Date: 2026-09-22T00:00:00+00:00 - Categories: LLMs & Agents - Authors: Terra Sztain - Journal: Nature Computational Science - DOI: 10.1038/s43588-026-01052-3 - Keywords: LLM, enzyme - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Machine Learned Interatomic Forces as Inference-Time Physical Guidance for All-Atom Diffusion Peptide Design - Source: ChemRxiv (preprints) - Date: 2026-09-22T00:00:00Z - Categories: Design de novo, Targets & Structures, ML Potentials - Authors: Biao Zeng, Youyi Song, Jinfeng Liu - DOI: 10.26434/chemrxiv.15009322/v1 - External ID: 10.26434/chemrxiv.15009322/v1 - Keywords: molecular generation, MLIP, MACE, denoising, diffusion model, generative models, receptor - Source URL: - Dashboard article: Abstract: Diffusion-based biomolecular generative models can efficiently explore peptidereceptor sequence-structure space, but the learned generative distribution does not explicitly represent the underlying atomistic potential-energy landscape. As a result, generated complexes may contain locally unfavorable atomic arrangements even when they remain plausible under the learned structural prior. Here, we develop an inferencetime physical guidance framework that couples machine learned interatomic potential (MLIP)-derived forces to the all-atom reverse diffusion trajectory of RFDiffusion3 for receptor-conditioned peptide design. Rather than applying energy minimization after generation, atomistic physical information is introduced before the joint sequencestructure trajectory termination. MLIP-derived forces therefore provide atomic coordinate corrections that are propagated through subsequent denoising steps, allowing physical information to influence not only atomic geometry but subsequent residue selection. The framework requires neither retraining nor modification of the pretrained diffusion model and uses late-stage force guidance, adaptive step-size control, and geometry-preserving projection to retain the underlying generative prior. Across eight receptor systems for peptide designs, MLIP guidance reduced peptidereceptor steric clashes by 24.7% and intrapeptide clashes by 37.4%, while promoting more hydrogen-bond-compatible interface geometries. Guided structures shifted toward more favorable energetic regions in both the guiding MACE potential and an independent MM/GBSA evaluation, with 82.4% of retained structures exhibiting more favorable MM/GBSA energetic estimates. These improvements were achieved with a median peptide backbone deviation of only 0.051 Å. Importantly, 95.0% of generated peptide sequences were preserved, whereas 5.0% underwent residue substitutions, demonstrating that physical guidance can feed back into sequence-structure generation without globally overriding the learned RFDiffusion3 distribution. This distinguishes inference-time physical guidance from post-generation relaxation, which optimizes only the coordinates at fixed chemical identity. More broadly, the results establish a general strategy for coupling pretrained biomolecular generators with differentiable atomistic energy models, enabling physical information to participate directly in molecular generation rather than serving only as a post hoc filter or refinement step. ## Machine Learning-Guided Screening and Experimental Validation of Plant-Derived Antifungal Compounds for Protecting Bamboo against Mold - Source: ACS Sustainable Chemistry & Engineering (journals) - Date: 2026-09-22T00:00:00Z - Categories: Property Prediction, ADMET & Safety - Authors: Jian-Liang Ding, Ying Ran, Quan Yuan, Hong-Tao Yu, Jia-Wei Zhu, Chun-Gui Du - Journal: ACS Sustainable Chemistry & Engineering - DOI: 10.1021/acssuschemeng.6c07168 - External ID: 3c8ff1568f65f27ef8edff27e94bea6cce525b15 - Keywords: QSAR, Molecular docking, XGBoost - Source URL: - Dashboard article: Abstract: Bamboo is a renewable material, but it is highly susceptible to mold growth under warm and humid conditions, which can significantly impair its serviceability and practical value. In this study, an XGBoost-based QSAR model integrating physicochemical descriptors and Morgan fingerprints was developed to screen plant-derived antifungal compounds against Aspergillus niger. The model achieved an R2 of 0.983 for the training set. Model-guided screening and subsequent antifungal assays identified vanillin, p-anisaldehyde, and geraniol as promising candidates. All three compounds inhibited common bamboo molds, and vanillin provided the best overall protection for sliced bamboo veneers. A source-expanded screening life cycle assessment showed that, with 80% ethanol recovery, the vanillin route had a GWP100 of 23.90 kg CO2-eq per 1000 m2 of treated sliced bamboo veneer, compared with 48.61 kg CO2-eq for ACQ-D, representing a reduction of 24.71 kg CO2-eq (50.83%). FTIR and SEM analyses indicated that impregnation altered the surface morphology and molecular environment of bamboo without producing readily detectable new functional groups. Physiological assays showed that all three compounds increased fungal membrane permeability and reduced ergosterol content. Molecular docking further suggested that the compounds may interact with the catalytic pocket of lanosterol 14α-demethylase (CYP51A), potentially interfering with ergosterol biosynthesis. These findings demonstrate the value of integrating machine-learning screening, experimental validation, mechanistic analysis, and environmental assessment to develop more sustainable mold-protection strategies for bamboo. ## Mature complexes carry more G protein class information than partner free GPCR structures - Source: ChemRxiv (preprints) - Date: 2026-09-22T00:00:00Z - Authors: Hossein Batebi - DOI: 10.26434/chemrxiv.15009240/v1 - External ID: 10.26434/chemrxiv.15009240/v1 - Keywords: GPCR, receptor - Source URL: - Dashboard article: Abstract: GPCR structures are widely used to infer G protein coupling and pathway selective pharmacology, but mature receptor G protein complexes may contain structural information acquired or stabilized after partner selection. We asked how much G protein class information is present independently of the bound transducer. We analyzed 1,360 experimental class A GPCR structures using structural bioinformatics, molecular recognition descriptors and receptor grouped machine learning. Every receptor with both a strict partner free active structure and a G protein bound structure entered the matched analysis; activation and evolutionary transfer served as controls. Twelve receptors met the strict criteria. Bound endpoints gave a larger G protein class margin in 11 of 12 receptors (mean difference 0.389, 95% CI 0.095 to 0.682; exact sign flip P = 0.030), with macro ROC AUC 0.795 versus 0.619. A four feature geometry classified activation at ROC AUC 0.996 across 1,336 structures from 199 receptors. Across 207 receptor transducer units, class information was distributed across sequence and structural properties, strongly shaped by evolutionary relatedness, and no universal cross family interface rule emerged. Mature GPCR complexes therefore contain receptor compatibility plus information associated with the selected and stabilized transducer. Similarity to a G protein bound endpoint supports compatibility with that signalling state but does not by itself establish intrinsic coupling preference, a distinction relevant to pathway selective drug discovery. ## ML-RKIM: Physics-Informed Machine Learning for Reaction Kinetics Identification and Modeling - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-22T00:00:00+00:00 - Categories: Cheminformatics - Authors: Paulina Portales Picazo, Navid Zobeiry - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c01936 - Source URL: - Dashboard article: Abstract: Kinetic processes play a central role in structural transformations across natural, engineered, and biological systems, ultimately determining their final properties. Traditional kinetic characterization methods are often time-consuming and resource-intensive, and require prior assumptions about the reaction mechanism. In this work, we present a physics-informed machine learning framework to extract closed-form kinetic representations from time-series data without preselecting a single reaction mechanism. The framework combines neural networks, physics-informed features, and sparse regularization. The framework was applied to a range of processes, including thermally driven reactions such as polymer curing and crystallization, and nonthermal processes such as enzymatic browning. The method identified sparse, closed-form kinetic representations while reducing reliance on a single predefined kinetic model. R2 values exceeded 98% for the synthetic cases and ranged from 76% to 92% for the experimental cases. In addition, a systematic ablation study evaluated robustness to noise, data availability, sampling density, incomplete measurements, and candidate function library variation, and an experimental case study was included to assess the performance under realistic conditions. These results demonstrate the potential of this approach to accelerate kinetic characterization across polymer, food, and biomolecular processing applications, where accurate kinetic models are essential. ## ML26: New Ingredients for an Integral-Features Density Functional with Broad Chemical Accuracy - Source: Journal of Chemical Theory and Computation (journals) - Date: 2026-09-22T00:00:00+00:00 - Authors: Dayou Zhang, Yinan Shu, Benjamin G. Janesko, Donald G. Truhlar - Journal: Journal of Chemical Theory and Computation - DOI: 10.1021/acs.jctc.6c01681 - Keywords: density functional theory, DFT - Source URL: - Dashboard article: Abstract: Integral-features density functional theory (IF-DFT) replaces the conventional integration over local ingredients with a nonlinear mapping from global integral descriptors, enabling machine-learned density functionals with broad transferability. Here we report ML26@MN15, a new IF functional that extends the earlier ML25@MN15 model by incorporating 12 additional correlation features─including overlap-projected rung-3.5 terms, CS1 dynamic-correlation ingredients, and VV10 nonlocal correlation─and by treating Hartree–Fock exchange as an integral feature rather than as a fixed hybrid percentage. ML26@MN15 employs 79 integral features evaluated on MN15 ingredients and a single-hidden-layer neural network with 5000 nodes to produce a size-extensive exchange–correlation energy. Trained on 185 databases comprising 7242 reference data, ML26@MN15 achieves a data-averaged energetic mean unsigned error of 0.89 kcal/mol, improving upon ML25@MN15 by 15% and outperforming a group of previous leading functionals across all eight chemical partitions examined. For the widely used MDB2019S, MGCDB84, and GMTKN55 data sets, ML26@MN15 yields the lowest averaged errors among all tested functionals, including the results for GMTKN55 by ML-based Skala-1.1 and DM21 density functional approximations. The new functional also maintains strong performance for systems containing heavy elements and exhibits improved behavior on a challenging self-interaction benchmark. These results show that expanding the integral-feature space with physically motivated nonlocal ingredients yields a density functional with unprecedented accuracy across diverse chemical benchmarks. The broad and consistent accuracy of ML26@MN15 highlights the promise of integral-features density functional approximations as a scalable framework for incorporating additional nonlocal physics into future functional development. ## ncAA-RepDistill: a chemistry-native SMILES representation for permeability prediction and local edit ranking of non-canonical cyclic peptides - Source: Journal of Cheminformatics (journals) - Date: 2026-09-22T00:00:00+00:00 - Categories: Cheminformatics, ADMET & Safety - Authors: Qiule Yu, Zhixing Zhang, Xiang Li, Weihua Li, Guixia Liu, Yun Tang - Journal: Journal of Cheminformatics - DOI: 10.1186/s13321-026-01311-5 - Keywords: SMILES, permeability prediction - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Solid-Phase Standard Enthalpies of Formation from SMILES with a Calibrated Per-Molecule Uncertainty: Application to a 13 606-Compound CHONFCl Inventory - Source: ChemRxiv (preprints) - Date: 2026-09-22T00:00:00Z - Categories: Cheminformatics, ML Potentials - Authors: Luc E. Brunet - DOI: 10.26434/chemrxiv.15009244/v1 - External ID: 10.26434/chemrxiv.15009244/v1 - Keywords: SMILES, ANI 2x, neural network potential - Source URL: - Dashboard article: Abstract: The standard enthalpy of formation of a solid, ∆fH ◦(s, 298 K), governs the detonation performance and the thermal hazard of an energetic material, yet it is measured for only a few hundred compounds, and 30 % of the compounds measured more than once carry determinations that disagree by more than 20 kJ mol−1. I report a route from a SMILES string to ∆fH ◦(s, 298 K) that returns, together with the value, a molecule-specific standard deviation calibrated against experiment. The gas-phase enthalpy is obtained from the ANI-2x neural network potential evaluated on a connectivity-based hierarchy (CBH-2, with CBH-1 fallback) isodesmic balance in which the nitro group is treated as a single pseudo-atom, without which the balance cannot be closed on a charge-separated nitrogen; a rigid-rotor harmonic-oscillator correction supplies the thermal terms. The sublimation enthalpy is a 32-descriptor additive model fitted to 229 measurements harvested from the NIST Chemistry WebBook. Two systematic shifts that survive leave-one-out validation are removed. The uncertainty is a heteroscedastic Gaussian variance model. The whole chain is validated by a nested leave-one-out procedure in which the sublimation model, the acceptance threshold, the systematic shifts and the variance model are all re-estimated without the compound they predict, and that compound’s own gas-phase reference is withheld: over 281 compounds the mean absolute error is 26.1 kJ mol−1, the mean |z| is 0.80 against 0.80 expected for a well-calibrated Gaussian, and the coverage is 71.2 % at one standard deviation and 93.6 % at two, against 68.3 % and 95.4 % expected. For scale, independent experimental determinations of the same solid deviate from their own median by 7.6 kJ mol−1 on average, so two determinations of one compound typically differ by about twice that. The procedure was applied to an inventory of 13 606 CHONFCl compounds; 13 267 (97.5 %) were predicted in 314 core-hours on a single desktop machine, the remainder being open-shell radicals outside the domain of ANI-2x. I state explicitly, and quantify, the size beyond which the uncertainty model itself extrapolates. ## Surfactant-Mediated Enantioselective Extraction: Interfacial Behavior Investigation from a Microscopic Perspective - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-22T00:00:00+00:00 - Categories: Free Energy & MD - Authors: Yunna Xue, Xinggui Zhou, Xiangyang Zhang - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c01740 - Source URL: - Dashboard article: Abstract: Enantioselective liquid–liquid extraction (ELLE) based on cyclodextrin host–guest recognition is a promising green strategy for chiral resolution. However, its industrial application is often hindered by restricted chiral recognition efficiency and ill-defined interfacial mass transfer mechanisms. In this work, a multiscale approach─integrating macro-equilibrium experiments, molecular dynamics simulations, and quantum chemical calculations─was employed to decode the interfacial molecular interactions and their decisive roles in the ELLE of ibuprofen. The interface was regulated and reconstructed through the introduction of surfactants. The anionic surfactant SDS shortened extraction equilibrium time by approximately 30%, while the facial amphiphilic NaDC extended equilibration but improved the enantiomeric excess. MD simulations revealed that the addition of SDS and NaDC increased the interfacial thickness from 0.41 nm by 1.23 and 1.51 times, respectively. Via umbrella sampling, the chiral migration process exhibited characteristic free-energy changes of approximately −11.48 kJ·mol–1 and +35.58 kJ·mol–1 when transferring from the organic and aqueous phases to the interface, respectively. Furthermore, quantum chemical analysis quantified the interaction energy differences, demonstrating that steric hindrance and hydrogen-bonding intensity variations drove the observed changes in equilibrium parameters. These findings establish a conceptual microscopic model for interface-mediated chiral partitioning, providing a robust theoretical framework for optimizing separation efficiency and process intensification in ELLE systems. ## SwinTransRNA: A Swin Transformer-Based Framework for Accurate Prediction of RNA Subcellular Localization - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-22T00:00:00+00:00 - Categories: Cheminformatics - Authors: Bowen Shi, Xuxin He, Yen-Peng Chiu, Zhihao Zhao, Peilin Xie, Yixian Huang, Xingchen Liu, Xiangrong Liu, Tzong-Yi Lee, Leyi Wei, Jiahui Guan, Jijun Tang, Lantian Yao - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c02004 - Source URL: - Dashboard article: Abstract: RNA subcellular localization determines the regulatory context in which microRNAs (miRNAs), circular RNAs (circRNAs), and long noncoding RNAs (lncRNAs) exert their functions, yet these RNA classes differ greatly in sequence length and circularity, making fixed-length raw-sequence representations inconvenient for cross-RNA comparison. Here, we present SwinTransRNA, a unified framework that encodes each RNA as a length-normalized 64 × 64 frequency chaos game representation (FCGR) of 6-mer composition, where every cell carries a fixed 6-mer identity and the matrix entries sum to one, enabling direct comparison across sequence lengths. A hierarchical window-attention network captures localized interactions among neighboring composition regions and enlarges the receptive field through shifted windows and patch merging. On audited benchmark data sets for nuclear/cytoplasmic lncRNA, nuclear/cytoplasmic circRNA, and intracellular/extracellular miRNA classification, SwinTransRNA achieves the highest accuracy and F1 among the compared methods on all three tasks (0.748/0.723, 0.914/0.908, and 0.819/0.823 for lncRNA, miRNA, and circRNA, respectively), surpassing RNALight and RNALoc-LM. Ablations show consistent gains over one-hot Transformer baselines and classical FCGR-based classifiers, while feature-space and curve-level analyses characterize the complementary contributions of the FCGR representation and the window-attention backbone. The main novelty lies in combining an order-traceable FCGR with hierarchical window attention under a reproducible protocol: five-seed repeats, confidence intervals, and paired tests quantify comparison uncertainty, and patch-embedding heatmaps with candidate motif screens expose the patterns prioritized by the model. The unified implementation, traceable sequence processing, and compact 4.25-million-parameter model provide a fixed-dimensional, auditable basis for RNA localization prediction and pattern prioritization. ## Synthesis, SAR Analysis, and Antifungal Potency against Plant Pathogenic Fungi of New Azidopyrazoles - Source: ACS Omega (journals) - Date: 2026-09-22T00:00:00+00:00 - Categories: Docking & Screening - Authors: Daniel Canseco-González, Rosa E. Sánchez-Fernández, Elizabeth Navarro-Cerón, Stephania Olivares-Sánchez, Alejandro O. Viviano-Posadas, Alejandro Dorazco-González, Ricardo Parra-Unda, Gilmar López-Armenta, Adrián L. Orjuela, Jorge Alí-Torres - Journal: ACS Omega - DOI: 10.1021/acsomega.6c07653 - Keywords: Molecular docking, IC50, binding affinity - Source URL: - Dashboard article: Abstract: The development of potent synthetic antifungal agents for plants is an ongoing challenge of modern chemistry that impacts biological, medicinal and environmental sciences. Herein, a concise Structure–Activity Relationship (SAR) model involving seven synthetic azidopyrazoles toward three pathogens of agricultural products and ornamental plants: Fusarium oxysporum, Rhizoctonia sp., and Alternaria alternata is described. The set of azidopyrazoles 1–7 includes a central fragment of 3-azido-1H-pyrazole substitutes in different positions of the ring (5-substituted, R = H, 1; R = −CH3, 2; R = –Ph, 3; R = –PhOCH3, 5; 4-substituted, R = –Ph, 4; R = –(4-Cl-Ph), 6) and 3-substituted 5-azido-1H-pyrazole (R = thiophene, 7) which are prepared via a high-yielding (>85%), one-step synthesis from commercial cheap aminopyrazoles. Compounds 3 and 7 showed the greatest inhibition with IC50 values of 26.52 ± 1.14 and 26.75 ± 1.36 μg/mL for F. oxysporum; 18.63 ± 1.08 and 14.39 ± 1.09 μg/mL for Rhizoctonia sp.; and 15.53 ± 1.14 and 21.02 ± 1.37 μg/mL for A. alternata. SAR analysis revealed that aryl substitution at C5 increased markedly the antifungal potency. In contrast, unsubstituted and alkyl-substituted pyrazoles (1, 2) had low activity. Combining 3 with Ag-nanoparticles (NPs) doubled its efficacy only against Rhizoctonia sp. (AgNPs, 10 nm, 100 ng/mL). From IC50 = 18.63 ± 1.08 to 9.65 ± 1.38 μg/mL). Molecular docking with MEP36 supported these results: compound 3 had the highest binding affinity (−6.8 kcal/mol) due to π–alkyl interaction with Val274 residue. Furthermore, root mean square deviation analyses confirmed the tight binding mode. ## The role of shape and interaction directionality in the crystalline phase behaviour of octahedral metal-organic cages - Source: Nature Communications (journals) - Date: 2026-09-22T00:00:00+00:00 - Categories: Free Energy & MD - Authors: Emma H. Wolpert, Andrew Tarzia, Yuhang Zhu, Kim E. Jelfs - Journal: Nature Communications - DOI: 10.1038/s41467-026-77857-7 - Keywords: molecular shape - Source URL: - Dashboard article: Abstract: Despite advances in crystal engineering, predicting the packing of metal-containing molecules remains a significant challenge due to their increased structural complexity and computational cost relative to purely organic compounds. Yet, predicting solid-state structures is critical for understanding structureproperty relationships, particularly in modular porous systems such as metal–organic cages (MOCs), where packing governs access to the internal pores and thereby impacts functionality. Here, we present a computationally inexpensive methodology that combines semi-empirical dimer calculations with coarse-grained modelling to predict the packing behaviour of MOCs. This approach enables a priori prediction of the solid state and allows insight into how molecular shape and intermolecular interactions influence the crystalline phase behaviour of MOCs. Our findings provide a foundation for the rational design of MOCs with tailored solid-state structures and properties. ## Therapeutic targeting of cancer-associated PKM2 mutants by shikonin via mutation-induced pocket recognition. - Source: Physical chemistry chemical physics : PCCP (journals) - Date: 2026-09-22T00:00:00Z - Categories: Docking & Screening, Targets & Structures - Authors: Anandita Mitra, Sandip Paul - Journal: Physical chemistry chemical physics : PCCP - DOI: 10.1039/d6cp02406h - External ID: 57eab9fcbf2769aee1125c430f7d33a02ff00aa2 - Keywords: Molecular docking, MD simulations, kinase, enzyme - Source URL: - Dashboard article: Abstract: Pyruvate kinase (PKM2) is a key glycolytic enzyme involved in cancer metabolic reprogramming and frequently harbors oncogenic missense mutations that alter its structural and functional dynamics. Here, we investigate the molecular basis of recognition of the anti-cancer drug shikonin in wild type (WT) and six PKM2 mutants. Molecular docking identified three mutants (R246S, P117L, and H464A) exhibiting favourable binding of shikonin within a newly identified pocket adjacent to the allosteric site, while the WT accommodated shikonin in the known active site pocket, consistent with previous literature reports. Subsequently, 1 µs MD simulations confirmed reduced conformational fluctuations, higher values and persistent residue contacts in these mutants, whereas the other three mutants (K367M, R399E, and R455Q) displayed disrupted pocket integrity, lower values and unstable ligand binding. Rg, PCA, Fpocket program and residue communication network analyses revealed that the favourable binding mutants maintain a compact pocket architecture stabilized through efficient short-range allosteric coupling between the mutation sites and pocket residues and druggability scores greater than 0.5, while weak-binding mutants show weakened long-range communication pathways. Interaction profiling further identified stable hydrogen bonding and stacking interactions supported by favourable interaction energies. Overall, this study demonstrates that PKM2 mutations dynamically remodel the binding pocket and allosteric communication to facilitate shikonin recognition, highlighting the newly identified pocket as a promising therapeutic site for future PKM2 targeted drug design. ## TrmD: A Promising Antibacterial Target and Its Small-Molecule Inhibitors - Source: Journal of Medicinal Chemistry (journals) - Date: 2026-09-22T00:00:00+00:00 - Authors: Zhiwei Meng, Di Xia, Xintong Zhang, He Zhu, Fengrun Zheng, Zihan Zhang, Zuhai Lei - Journal: Journal of Medicinal Chemistry - DOI: 10.1021/acs.jmedchem.6c01066 - Keywords: virtual screening - Source URL: - Dashboard article: Abstract: Bacterial tRNA-(N1G37) methyltransferase (TrmD), a member of the SPOUT superfamily, plays a pivotal role in catalyzing the methyl group transfer from S-adenosyl-l-methionine to the N1 position of G37 in the pathway of post-transcriptional modifications of bacterial tRNA. In response to the urgent global clinical need to overcome antibiotic resistance, inhibiting TrmD activity has emerged as a novel strategy for combating various antimicrobial-resistant bacterial strains. Various rational drug discovery strategies, including fragment-based drug design, high-throughput screening, click chemistry-based direct-to-biology screening, virtual screening, molecular hybridization, and prodrug strategy, have been successfully employed to identify small-molecule TrmD inhibitors with diverse scaffolds. Here, we highlight discovery strategies, molecular structural features, and key structure−activity relationships that have driven the discovery and development of small-molecule TrmD inhibitors; discuss the key factors underlying the gap between enzymatic inhibitory potency and in vitro antibacterial activity of small-molecule TrmD inhibitors; and offer insights for bridging it. ## Uni-Macro-FRPN: Full-Resolution and Cross-Scale Learning for Polymers - Source: ChemRxiv (preprints) - Date: 2026-09-22T00:00:00Z - Authors: Jintao Wu, Yiran Shan, Rui Zhang - DOI: 10.26434/chemrxiv.15009049/v2 - External ID: 10.26434/chemrxiv.15009049/v2 - Keywords: Transformers - Source URL: - Dashboard article: Abstract: Polymer properties emerge from interactions across scales, yet existing polymer models typically preserve either detailed monomer chemistry without an explicit polymer graph or polymer connectivity with simplified monomer representations, due to computational constraints, as polymers typically contain tens of thousands of atoms. We present Uni-Macro-FRPN (FRPN), a Full-Resolution Polymer Network that retains both detailed atom-level and monomer-level features and explicit polymer structure information within a unified framework. Two Transformers jointly learn atom-informed monomer semantics, sequence order, and chain topology from BigSMILES-derived representations. On the Block Copolymer Database (BCDB) lamellar-versus-non-lamellar classification task, FRPN achieves 86.4% accuracy and 90.6% ROC–AUC, establishing state-of-the-art performance. Ablation results indicate that the gain is not explained solely by increased parameter count. On a linear homopolymer benchmark, monomer-centric learning remains competitive, highlighting a boundary case where polymer-scale organization is simple. To test generalization beyond linear polymers, we further construct an all-atom molecular-dynamics benchmark of 1640 datapoints spanning diverse monomer chemistries, sequence orderings, chain topologies, and physical properties. FRPN achieves the strongest overall performance on this topology-rich benchmark, with diagnostics supporting the benefit of jointly modeling monomer chemistry and polymer structure. Taken together, FRPN provides a practical route for moving polymer representation learning beyond monomer-centric representations. The leading performance of FRPN also suggests a promising direction for polymer informatics: future polymer prediction models should treat polymers not only as collections of monomer descriptors, but as complete multiscale chemical and topological objects. ## Unraveling a Case Where the NAC Approach Fails─MD and QM Modeling of Regioselective cis -Stilbene Oxide Hydrolysis by Engineered Epoxide Hydrolases - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-22T00:00:00+00:00 - Categories: Cheminformatics - Authors: Elvira Bombino, Hesam Arabnejad, Baoyan Liu, Peter A. Jekel, Xiang Sheng, Dick B. Janssen, Fahmi Himo, Hein J. Wijma - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c02029 - Source URL: - Dashboard article: Abstract: Molecular dynamics (MD)-based methods to predict the enantioselectivity or product specificity of enzymes typically use a near attack conformation (NAC) approach. The computational efficiency of this ground-state approach, taking minutes to hours, allows to in silico screen thousands of enzyme variants. During the computational design and screening of new variants of limonene epoxide hydrolase for improved stereoselectivity, it was noticed that, while this approach gave satisfactory prediction accuracy for most substrates, the enantioselectivity predictions were far less accurate for variants designed for cis-stilbene oxide. Here, to investigate whether inaccurate predictions with the NAC-based approach were typical for this substrate, first, a larger data set was generated by experimentally testing and computationally predicting the enantioselectivity of an additional 18 enzyme variants for cis-stilbene oxide. Subsequent MD simulations confirmed that a NAC analysis resulted in inaccurate enantioselectivity predictions throughout the data set. Density functional theory (DFT) calculations using the B3LYP-D3(BJ) functional, with an active site cluster model of >300 atoms, provided a rationale for why the NAC approach gave poor predictions with cis-stilbene oxide. The aromatic rings of the substrate influence the nucleophilic attack on the epoxide group, by providing a blend of stereoelectronic activation and steric hindrance to the transition state, which is not considered in the NAC approach. With this substrate, the DFT accurately predicted the enantioselectivity. The results offer valuable insights into the applicability of the NAC approach for predicting enzyme selectivity, suggesting that more expensive methodologies, which do explicitly consider the transition state structure, can also become required when bulky or stereoelectronically activating groups are bound to the reacting atoms. ## Unraveling the PFAS Helix: A Statistical Approach - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-22T00:00:00+00:00 - Categories: Property Prediction - Authors: Pranoy Ray, Haden Cavalli, Gashaw Bizana, Andrew R. Castillo, Shubham Vyas, Ronald L. Siefert, Surya R. Kalidindi, Manoj Kolel-Veetil - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c01874 - Source URL: - Dashboard article: Abstract: The extreme persistence of per- and polyfluoroalkyl substances (PFAS) in the environment is rooted in their three-dimensional molecular conformation. The helical twist adopted by perfluoroalkyl chains due to hyperconjugation involving their recalcitrant C–F bonds governs their resistance to degradation, yet a quantitative, continuous metric for helicity has remained absent. Here we introduce a data-driven framework that quantifies backbone helicity using three statistical descriptors: void-state (binary occupancy) spatial autocorrelations, local backbone principal component analysis, and persistent homology. We further validate each statistical descriptor against geometric dihedral benchmarks and DFT-calculated Vibrational Circular Dichroism (VCD) spectra. The framework is initially established on a homologous series of perfluorocarboxylic acids (FC2 through FC16) and their hydrogenated analogues, then extended across perfluorosulfonic acids, fluorotelomer alcohols, polyfluoroalkyl hexanoic acid analogues, and longer-chain PFOA and PFOS analogues. Agreement between statistical, geometric, and spectroscopic definitions of helicity is established for PFCAs and extended to structurally distinct subfamilies, including PFOA analogues of varying fluorine content and perfluorosulfonic acids, demonstrating that the descriptors are robust to changes in headgroup chemistry and fluorine substitution pattern. The framework provides a chemistry-agnostic toolset for incorporating backbone conformation into predictive models of PFAS environmental fate and degradation reactivity. ## ZeoAgent: Autonomous Design of Zeolite Frameworks through Pore-Topology-Guided Generation and Evaluation - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-22T00:00:00+00:00 - Categories: Property Prediction - Authors: Jing Ping, Zhendong Liu - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c02058 - Source URL: - Dashboard article: Abstract: Autonomous agents are emerging as a promising approach to assist and accelerate materials design and discovery. For zeolites, an important class of crystalline porous materials, this paradigm could address a long-standing challenge in framework discovery: the lack of rational strategies for coordinating framework generation with property objectives. Yet building such an autonomous agent system for zeolites remains challenging, as performance is governed by periodic pore topology and strict framework connectivity constraints. This requires the agent to jointly reason over pore architectures and generate valid frameworks. Here we present ZeoAgent, a zeolite-specific system that converts natural-language objectives into computational operations for zeolite structure analysis and property prediction as well as framework design. ZeoAgent uses a point-cloud representation of pore architecture to learn from existing zeolite frameworks and guide the generation of framework candidates, which are then evaluated against the design objective within the same loop. We demonstrate ZeoAgent in representative design tasks, including the design of frameworks with prescribed structural constraints, enhanced diffusion performance, and pore dimensions suited for CO2/CH4 separation. This work establishes a route for goal-directed zeolite design, showing how AI agents can be adapted to periodic porous materials governed by constrained pore topology. ## Spectra: A Rules-Driven LLM Pipeline for Automated KYC Document Processing - Source: arXiv (preprints) - Date: 2026-09-21T22:59:33Z - Categories: LLMs & Agents - Authors: Miray Wahib, Ethan Tran, Rea Mourad, Mira Muti, Nikita Dvornik - External ID: 2609.25474v1 - Keywords: LLM - Source URL: - Dashboard article: - PDF: Abstract: Know Your Client (KYC) onboarding in capital markets requires analysts to manually classify documents, extract structured data from heterogeneous sources, and validate compliance against complex regulatory policies. This process requires significant analyst time per client, with end-to-end onboarding often stretching to multiple weeks due to sequential handoffs. In this work, we analyze an on-boarding process and find that it comprises repeatable components well-suited to AI automation. We therefore propose a restructured workflow to be amenable to automation: we consolidate the traditional four-party process into two parties that share most of the work and can be automated together, eliminating intermediate handoffs that compound delays. To automate the remaining steps, we introduce Spectra, an AI-assisted document processing platform that combines a structured rules engine with LLM-based classification, extraction, and validation agents. The rules engine encodes compliance policy as a queryable database, enabling focused context injection that reduces token usage while improving extraction precision. Rather than a single monolithic prompt, the system decomposes document processing into isolated, auditable stages, each optimized independently and traceable to specific policy clauses. In evaluation on real KYC documents, Spectra achieves 100% classification accuracy and 89.4% extraction accuracy. Human review burden dropped by 96%. ## MolExplain: An Interactive Tool for Explainable Molecular Property Prediction - Source: arXiv (preprints) - Date: 2026-09-21T19:46:45Z - Categories: Property Prediction, ADMET & Safety - Authors: Pirm Dhararag, Dylan Cashman - External ID: 2609.25355v1 - Keywords: Property Prediction, XGBoost, Molecular Property - Source URL: - Dashboard article: - PDF: Abstract: The application of machine learning to molecular property prediction has become increasingly prevalent in drug discovery, yet most models operate as black boxes, returning a prediction without revealing which structural features drive it. MolExplain addresses this gap by combining property prediction with sub-structure level visual explainability in an interactive web interface. The system featurizes molecules as Morgan fingerprints, classifies them using a trained XGBoost model, and applies SHAP attribution to produce a smooth heatmap overlay indicating which regions of the molecule contribute for or against the predicted property. Applied to cyclic peptide membrane permeability, the tool's attribution independently recovers the known role of backbone N-methylation in improving passive membrane diffusion, consistent with established chemistry. While demonstrated on cyclic peptides, the framework is designed to generalize to other molecular properties, positioning MolExplain as a platform for interactive, explainability-driven molecular design. ## Exactness at Inference: A Representational Criterion for Out-of-Distribution Generalization - Source: arXiv (preprints) - Date: 2026-09-21T17:39:51Z - Authors: Filipe Marinho Rocha, Inês Dutra, Vítor Santos Costa, Luís Paulo Reis - External ID: 2609.24942v1 - Keywords: equivariant - Source URL: - Dashboard article: - PDF: Abstract: A model generalizes outside its training distribution only when it computes a representation structurally equivalent to the generating mechanism, not an approximation fitted to it. Such equivalence is necessary for exactness in and out of distribution, and extrapolation is governed by this exactness at inference, whatever its realization. Tensor Logic shows this: a zero-temperature contraction is equivalent to discrete logic, deducing in place with no artefact extracted, its tensors Boolean, its embeddings orthonormal, only its arithmetic continuous. Lacking infinite recursion it reaches Datalog, not Prolog, and though exact over closed domains it needs external memory to bind a novel entity. The criterion needs neither a discrete representation nor an extracted expression, and constrains inference, not training: an exact marginal in $\[0,1\]$ passes, a Neural Network thresholded to a hard label does not. Logic Tensor Networks fail it, while differentiable ILP and Tensor Logic at $T=0$ pass. Piecewise-affine extrapolation divergence and an inability to bind novel entities are two faces of a shortfall in exact representability. For hybrid architectures, a propagation rule follows: the output inherits the bounds of every fitted estimator on its path, explaining which axes fail in equivariant models and the ARC-AGI induction/transduction split. Only an exact hypothesis class certifies what the training data leave underdetermined: on a law-derived partition it finds the $56.3\\%$ of distant queries that are answerable, which ensembles meet with false confidence and distance metrics rank backwards. Common inductive biases, from symmetries to memory, reach exactness only because humans inject them, an argument for inducing exact representations rather than fitting surrogates whose residuals, even at the arithmetic floor in training, diverge outside the data and compound under composition. ## An accessible property classification framework to predict the solubility of functionalised naphthalenes and rylenes in organic solvents - Source: Digital Discovery (journals) - Date: 2026-09-21T16:23:14Z - Authors: Connor R.M. MacDonald, Alex Samuel Loch, Neil Berry, Emily Rose Draper - Journal: Digital Discovery - DOI: 10.1039/d6dd00275g - Source URL: - Dashboard article: Abstract: The development of functional materials such as multilayer films from supramolecular self-assembling materials is hindered by the current lack of design principles and predictability. Quantitative structure-property relationships have emerged as... ## Colour me shocked: Exact Molecular Hessians from local MLIPs in O(N) time using sparse differentiation! - Source: arXiv (preprints) - Date: 2026-09-21T14:56:56Z - Categories: ML Potentials - Authors: Luca Thiede, Andreas Burger, Alán Aspuru-Guzik - External ID: 2609.24720v1 - Keywords: MLIP - Source URL: - Dashboard article: - PDF: Abstract: The Hessian of the energy with respect to the nuclear positions is indispensable in atomistic modelling. However, constructing this matrix requires $O(N)$ Hessian vector products, traditionally limiting high-accuracy Hessians to small systems. Machine learning interatomic potentials (MLIPs) have accelerated atomistic modelling by providing highly accurate energies and forces at $O(N)$ cost, yet the resulting $O(N^2)$ cost of Hessians remains a practical bottleneck for large systems. Based on the insight that we can derive the sparsity pattern for an MLIP's Hessians in closed form, we show in this paper how to use techniques from sparse automatic differentiation to reduce the cost of a local MLIP's Hessians to a system-size-independent number of Hessian-vector products, yielding overall $O(N)$ total cost without any approximations. We benchmark our approach on a variety of systems ranging from alkane chains to water clusters to $A\\beta40$ conformers. Depending on the MLIP configuration, we achieve the linear scaling regime already on relatively small systems, resulting in large runtime reductions between 2$\\times$-15$\\times$ for these systems. This opens up the possibility of scaling high-accuracy MLIP Hessians to very large systems, such as proteins that were previously inaccessible. ## Composition-Dependent Self-Diffusion Coefficients in Liquid Mixtures from Hybrid Machine Learning - Source: arXiv (preprints) - Date: 2026-09-21T13:53:44Z - Categories: Cheminformatics, Property Prediction - Authors: Jens Wagner, Thomas Specht, Hans Hasse, Fabian Jirasek - External ID: 2609.24599v1 - Keywords: SMILES - Source URL: - Dashboard article: - PDF: Abstract: Self-diffusion coefficients are key descriptors of molecular mobility, yet experimental data remain scarce, highlighting the need for reliable prediction methods. In previous work, we introduced the hybrid Enhanced Stokes-Einstein (ESE) model, which advanced the state of the art in the physically consistent prediction of self-diffusion coefficients of solutes at infinite dilution in pure solvents by integrating the Stokes-Einstein equation with machine learning (ML). Here, we extend this approach to concentration-dependent self-diffusion coefficients and multicomponent solvents with HADES. This hybrid architecture leverages a deep-set neural network to connect pure-component and mixture prediction within a single framework. HADES predicts self-diffusion coefficients in liquid mixtures with any number of components at any composition and temperature. The only required inputs are SMILES-encoded molecular structures of the components and the pure-component viscosities, making the method broadly applicable. Trained and evaluated on a comprehensive dataset of 2526 data points for 600 systems, HADES significantly outperforms benchmark prediction methods. The trained model and its source code are fully disclosed, and the application is available via an interactive website https://ml-prop.mv.rptu.de/. ## QLoRA Fine-Tuning of Ministral LLM for Sequence-to-Function Protein Annotation - Source: arXiv (preprints) - Date: 2026-09-21T13:09:59Z - Categories: LLMs & Agents - Authors: Demian Pavlyshenko, Bohdan Pavlyshenko - External ID: 2609.24538v1 - Keywords: LLM, LLMs, GPT - Source URL: - Dashboard article: - PDF: Abstract: Functional annotation of newly sequenced proteins remains a bottleneck in molecular biology: the number of sequences in public repositories grows far faster than the capacity for manual curation. Most computational approaches consider annotation as multi-label classification over a fixed ontology, which constrains predictions to a predefined label set. In this work we study the the protein annotation as a sequence-to-text generation problem. We fine-tune the 3B-parameter Ministral 3 base model with QLoRA (4-bit NF4 quantization with low-rank adapters) on sequence annotation pairs. We assess predictions with an LLM-as-expert protocol: a GPT model prompted as a senior molecular-biology curator scores organism identification as binary and function annotation quality. We conclude that QLoRA-fine-tuned compact LLMs can generate curator-style annotations with genuine biological value for a substantial subset of proteins. We also discuss future directions in data quality, model scaling, and evidence grounding that are needed to make the approach sufficiently reliable for practical use. ## Pharmacokinetic State Space Models for Unbiased Prediction of Haemodynamic Collapse - Source: arXiv (preprints) - Date: 2026-09-21T09:35:19Z - Categories: ADMET & Safety - Authors: Rithin Nagaraj, Sudiksha Chindula, Bhaskarjyoti Das - DOI: 10.1007/978-3-032-35387-0\_22 - External ID: 2609.24338v1 - Keywords: Transformers, Pharmacokinetic - Source URL: - Dashboard article: - PDF: Abstract: An Intraoperative Hypotension (IOH) event is a frequent complication during administration of general anaesthesia with serious downstream consequences, yet clinical management remains reactive and not predictive. Existing predictive models, however, ignore drug infusion history as a valuable signal for prediction despite its direct pharmacological relevance. Our model achieves an Area Under the Receiver Operating Characteristic curve (AUROC) of 0.7360 and an Area Under the Precision-Recall Curve (AUPRC) of 0.1794, representing a 2.73-fold lift over the random guessing AUPRC baseline (0.0657), with the removal of propofol and remifentanil effect-site concentrations resulting in a 13.9% AUPRC drop compared to the full model. This is consistent with the hypothesis that pharmacokinetic trajectories encode impending haemodynamic changes before they manifest in the Mean Arterial Pressure (MAP). Additionally, this paper shows that training without lead-gap filtering degraded AUROC by 16.7%, empirically confirming that unfiltered models learn to detect ongoing hypotension rather than predict future events. Finally, a Mamba-based architecture achieves the aforementioned high prediction performance while maintaining a constant memory footprint across a range of sequence lengths, unlike the quadratic VRAM overhead typical of vanilla Transformers, making it the more practical choice for continuous intraoperative deployment. ## Exploring Conformers of Small Rings: Small Experiments and Observations \#RDKit \#cheminformatics \#memo - Source: Is life worth living? (iwatobipen) (feeds) - Date: 2026-09-21T09:17:41+00:00 - Categories: Blog - Keywords: RDKit, cheminformatics - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Adapting Boltz-2 with limited experimental activity data improves early enrichment in virtual screening - Source: arXiv (preprints) - Date: 2026-09-21T09:02:04Z - Categories: Docking & Screening, Targets & Structures - Authors: Kairi Furui, Masahito Ohue - External ID: 2609.24302v1 - Keywords: virtual screening, Boltz 2 - Source URL: - Dashboard article: - PDF: - Code: Abstract: Virtual screening aims to prioritize active compounds from large chemical libraries within a limited experimental budget. When applying Boltz-2 to virtual screening, a key challenge is how to use limited experimental data from the target assay to improve the prioritization of active compounds. We investigated whether fine-tuning the Boltz-2 affinity heads with a small number of binary activity labels could improve early enrichment of active compounds in hit discovery. We compared fine-tuning with 40-300 labels in a retrospective evaluation on eight MF-PCBA targets. With 300 activity measurements, fine-tuning increased the number of actives in the top 1% by a geometric mean of 1.77-fold across the eight targets and improved average precision (AP) by 2.14-fold relative to the control without fine-tuning. We also investigated whether rescoring a subset of candidates could retain the improvement in hit recovery by reranking only the top-ranked Boltz-2 candidates with the fine-tuned head. Restricting rescoring to approximately 10% of the evaluation set retained hit recovery comparable to full rescoring. These findings show that affinity-head fine-tuning with limited activity labels improves early enrichment with Boltz-2 and that this benefit can be retained when rescoring a restricted set of candidates. ## From Heuristics to Machine Learning: The Performance Ceiling for Single-Ion Magnets and Its Electronic Origin - Source: arXiv (preprints) - Date: 2026-09-21T03:06:12Z - Categories: Property Prediction - Authors: Federico Zahariev, Regina Pereyra, Vassiliki-Alexandra Glezakou, Durga Paudyal - External ID: 2609.24038v1 - Keywords: ab initio - Source URL: - Dashboard article: - PDF: Abstract: Machine learning (ML) is expected to speed up the discovery of single-ion magnets (SIMs), but does the structural information available before synthesis allow such predictions? For 1215 lanthanide complexes from the SIMDAVIS 1.2.1 database we compared three increasing levels of structural description: tabular features of the coordination site, continuous symmetry measures of the coordination polyhedron, and the complete 3D arrangement of atoms. All three converge to an accuracy near 76%, only slightly above the 71% of the single rule "predict SIM for Dy3+". To explain the failures, we combined multireference ab initio calculations with an inspection of the structures behind the high-confidence errors. The SIMs missed by the geometric models are field-induced relaxers whose ground Kramers doublets are prone to tunnelling, a property invisible to geometric descriptors. Many false positives contain several lanthanide centers or radicals, so their relaxation is collective and outside the single-ion picture. The electronic-structure and connectivity information needed to identify SIMs is therefore not accessible to geometric methods alone. Geometric models remain useful: restricting the screening to compounds with high prediction confidence raises the accuracy to 88% while retaining 48% of the dataset. Building on the analysis of the failures, we propose a strategy that combines simple filters for nuclearity and for radicals with ligand-field descriptors from ab initio calculations. ## UniK: Universal Knowledge Perception for Digital and Physical AI - Source: arXiv (preprints) - Date: 2026-09-21T00:56:36Z - Categories: LLMs & Agents - Authors: Nirmit Desai, Kunal Sawarkar, Aditya Mahakali, Dongkon Lee, Kevin Park, Eric Song - External ID: 2609.23971v1 - Keywords: LLMs, GPT - Source URL: - Dashboard article: - PDF: Abstract: Two transformative classes of AI systems are reshaping how organizations operate: \\textit\{digital AI\}, which reasons over enterprise knowledge to power chatbots and agent workflows; and \\textit\{physical AI\}, which learns to control robots and autonomous systems from video, gameplay, and sensor telemetry. Both face the same foundational bottleneck: raw knowledge at scale, spanning heterogeneous modalities, locked in private corpora that existing AI infrastructure cannot access reliably or efficiently. We propose \\textit\{Universal Knowledge Perception (UniK)\} as a common platform for both classes, covering the full knowledge lifecycle (ingestion, enrichment, indexing, retrieval, and continuous evaluation) across modalities from rich text and video to molecular data and sensor telemetry. We present UniK, built on Polymath Retrieval (multi-index fusion over automatically enriched indices) with no task-specific fine-tuning. Across five digital AI domains (medical literature, open-domain QA, chemistry, legal video proceedings, and government open data) UniK combined with an open-source 70-billion-parameter model consistently matches or outperforms frontier proprietary LLMs that are orders of magnitude larger: 76\\% RAG accuracy on government data versus 47\\% for GPT-5; 77.9\\% on medical QA without fine-tuning; topping all open-source chemistry pipelines. We show that the same infrastructure directly addresses the data curation, indexing, and retrieval challenges facing physical AI world model training, where the knowledge problem is harder but structurally identical. ## A machine-learning-driven dataset of 144,111 polycyclic aromatic hydrocarbon infrared spectra for interpreting aromatic infrared bands - Source: Astronomy & Astrophysics (journals) - Date: 2026-09-21T00:00:00Z - Authors: Xing-Hong Mai, Zu-Jia Lu, Zhao Wang - Journal: Astronomy & Astrophysics - DOI: 10.1051/0004-6361/202662240 - External ID: 1a34fbcff65870e2a1aa1fa19f65516d821435c7 - Keywords: density functional theory, force fields - Source URL: - Dashboard article: Abstract: The interpretation of interstellar aromatic infrared bands (AIBs) relies on reference spectral libraries, but the chemical space covered by existing databases remains incomplete relative to the vast structural diversity of polycyclic aromatic hydrocarbons (PAHs) expected in the interstellar medium. We present a large-scale, high-throughput PAH infrared (IR) spectral database that complements existing reference libraries by providing significantly more continuous coverage of the PAH chemical space for small to medium-sized PAHs. We used a machine-learning double-harmonic (ML-DH) framework that integrates neural network force fields with electronic dipole models. This framework achieves quantum-chemical accuracy at a fraction of the computational cost. It has been validated against independent density functional theory calculations and benchmark spectral databases. The resulting dataset comprises 144,111 IR spectra for 48,037 closed-shell, even-carbon benzenoid PAH structures across neutral, cationic, and anionic charge states, complete with per-spectrum uncertainty estimates. The spectral library, optimized geometries, and model code are publicly available. They can be used to complement existing libraries for interpreting and modeling Space Telescope (JWST) AIB observations and those from future IR missions. James Webb ## A Multistage Computational Framework for Early Prioritization of Natural Anti‐Inflammatory Candidates From Diverse Chemical Spaces - Source: ChemMedChem (journals) - Date: 2026-09-21T00:00:00+00:00 - Categories: Docking & Screening, ADMET & Safety - Authors: Huynh Anh Duy, Tarapong Srisongkram - Journal: ChemMedChem - DOI: 10.1002/cmdc.70497 - Keywords: virtual screening, molecular docking, toxicity prediction - Source URL: - Dashboard article: Abstract: Identifying safe and effective anti‐inflammatory agents is challenging, especially given the adverse effects of long‐term conventional therapies. Here, we present a multistage computational framework to prioritize early natural compounds with potential anti‐inflammatory activity across diverse chemical spaces. The proposed workflow integrates machine learning, consensus prediction, drug‐likeness evaluation, synthetic accessibility assessment, toxicity prediction, and molecular docking to enable stepwise refinement of candidate selection. Six predictive models were developed and evaluated with gold‐standard metrics, all showing consistent performance with accuracies, AUROCs, and AUPRCs above 0.8. A consensus strategy was used to reduce model variability and improve prediction reliability. SHAP analysis identified structural scaffolds associated with predicted anti‐inflammatory activity, providing interpretability for compound prioritization. Sequential filtering selected compounds that met predefined criteria for drug‐likeness, synthetic feasibility, and predicted safety profiles. Application of the framework to the NPASS database yielded 20 prioritized natural products, of which 17 candidates were further selected based on molecular docking analysis, exhibiting predicted binding affinities of <−7 kcal/mol. This framework was deployed on a web server, CAIP ( https://caip‐predictor.streamlit.app/ ), enabling rapid prediction and preliminary virtual screening of novel compounds. In total, the proposed framework offers a computational approach to systematically support the prioritization of anti‐inflammatory candidate molecules. ## A Systematic Evaluation of Molecule Generation Models for De Novo Drug Design: From Benchmarks to Practical Insights - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-21T00:00:00+00:00 - Categories: Cheminformatics, Property Prediction, Docking & Screening, Design de novo - Authors: Xinrui Xu, Xueer Wang, Dan Luo, Sisi Yuan, Xuan Lin - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c02405 - Keywords: Molecule Generation, de novo drug design, virtual screening, diffusion models, transformer, recurrent neural, RNN, molecular representations, receptor - Source URL: - Dashboard article: - Code: Abstract: Molecule generation has emerged as a powerful computational tool for de novo drug design, enabling the exploration of the chemical space beyond the limits of conventional virtual screening. The field has progressed rapidly, driven by advances in molecular representations, generative architectures, and target-aware modeling strategies. However, existing reviews typically address specific model families or application scenarios in isolation rather than offering an integrated perspective on how these components collectively form a coherent generation workflow. In this review, we present a comprehensive evaluation of molecule generation models for de novo drug design, covering 82 methods across five deep generative frameworks, including recurrent neural network (RNN)- and transformer-based models, variational autoencoders (VAEs), generative adversarial networks (GANs), flow-based models, and diffusion models. We first summarize widely used benchmarks and molecular representations and then examine the methodological principles underlying both general and pocket-conditioned generation. A central contribution of this work is a systematic synthesis and comparative analysis of the reported performance across commonly used benchmarks and evaluation metrics. We also summarize representative experimentally validated cases. Looking ahead, we discuss future directions in standardized 3D data, interaction-aware generation, receptor flexibility, and multiobjective molecular design, with the aim of improving the reliability and experimental relevance of molecule generation. All collected benchmark resources, evaluation metrics, and model references are provided in a publicly accessible repository at https://github.com/JacklinGroup/molecule-generation-review. ## A theoretical and empirical evaluation of product connectivity descriptors for improving antiviral QSAR and QSPR prediction - Source: BMC Bioinformatics (journals) - Date: 2026-09-21T00:00:00Z - Categories: Property Prediction - Authors: M. Alsharafi, Azzam Altairi, Zaied Alhaj, Yusuf Zeren - Journal: BMC Bioinformatics - DOI: 10.1186/s12859-026-06654-2 - External ID: 71e6b5f3eedda0c9bbb90a75d65a371580a74c92 - Keywords: QSAR, QSPR - Source URL: - Dashboard article: - Abstract: not stored for this record. ## AI-guided ethnopharmacology for cardiovascular drug discovery: from biomedical data to experimental validation - Source: Frontiers in Pharmacology (journals) - Date: 2026-09-21T00:00:00Z - Categories: Property Prediction - Authors: Shun Yao, Xiao-Xin Chen - Journal: Frontiers in Pharmacology - DOI: 10.3389/fphar.2026.1962508 - External ID: 957d67a801865f65960f80dbf4b9e1eb5ce47fbc - Keywords: property prediction, molecular property - Source URL: - Dashboard article: Abstract: Despite substantial advances in pharmacological therapy, cardiovascular disease remains the leading cause of mortality worldwide. Botanical drugs represent a rich source of structurally diverse bioactive metabolites, yet their translation into modern cardiovascular therapeutics remains limited by chemical complexity, incomplete mechanistic understanding, and challenges in experimental validation. Meanwhile, the rapid expansion of biomedical data resources and advances in artificial intelligence (AI) have created new opportunities to systematically investigate botanical drugs beyond empirical approaches. In this review, we summarize the major biomedical data resources supporting AI-guided ethnopharmacology, including phytochemical and ethnomedicinal databases, multiomics datasets, single-cell and spatial transcriptomics, perturbation-response resources, biological interaction networks, and clinical evidence. We then discuss the computational approaches used for molecular property prediction, drug-target interaction prediction, perturbation-response modeling, and foundation-model transfer learning, highlighting how these complementary strategies address distinct stages of pharmacological discovery. Building on these advances, we propose two complementary AI-guided discovery frameworks: a compound-first strategy that predicts the pharmacological activities of botanical drugs and a disease-first strategy that identifies natural products capable of reversing disease-associated molecular and cellular states. We further discuss emerging applications in cardiovascular diseases, emphasizing the integration of AI with single-cell atlases, disease-state prediction, mechanistic investigation, and experimental validation to prioritize candidate botanical therapeutics. Finally, we outline current challenges in data standardization, model robustness, interpretability, and prospective validation. We propose that AI-guided ethnopharmacology should be viewed as a hypothesis-generation framework that complements, rather than replaces, pharmacological experimentation. When supported by standardized botanical materials, high-quality multimodal data, rigorous biological validation, and prospective clinical evaluation, AI has the potential to accelerate cardiovascular botanical drug discovery and facilitate the translation of ethnomedicinal knowledge into modern precision therapeutics. ## AIDrugDesigner: a web server for drug-like molecules generation and optimization - Source: Bioinformatics (journals) - Date: 2026-09-21T00:00:00+00:00 - Authors: Huimin Zhu, Renyi Zhou, Shiliang Zhang, Yifan Wu, Zhangli Lu, Dongsheng Cao, Min Li - Journal: Bioinformatics - DOI: 10.1093/bioinformatics/btag693 - Keywords: drug likeness, LogP - Source URL: - Dashboard article: Abstract: Motivation Designing molecules with predefined properties remains a critical yet challenging task in drug discovery. Recent advances in deep learning have demonstrated strong potential in accelerating the identification of novel compounds with desired properties and activities. However, the implementation and running of these models often present difficulties due to varying environments and the requisite for additional coding competencies. Results Here, we present AIDrugDesigner, an integrated web server for practical deep learning–driven drug design. The platform consists of two modules: generation and optimization. The generation module supports target-aware drug design, pharmacophore-guided, property-conditioned, and linker design from diverse inputs. The optimization module refines user-provided molecules with respect to drug-likeness, LogP, and synthetic accessibility. By integrating multiple generative strategies into a unified interface, AIDrugDesigner reduces the technical barrier for applying deep learning models in drug discovery. Systematic evaluations across multiple design tasks further demonstrate its versatility. Availability and implementation The server is freely available at https://www.csuligroup.com/AIDrugDesigner. ## Antimicrobial Resistance Breakers That Specifically Restore Efficacy of Polymyxins - Source: ChemRxiv (preprints) - Date: 2026-09-21T00:00:00Z - Authors: Beth A. Mullan, Catherine R. Webley, Prachi Bendale, Miguel A. Valvano, Gerd K. Wagner - DOI: 10.26434/chemrxiv.15009183/v1 - External ID: 10.26434/chemrxiv.15009183/v1 - Keywords: in silico screening, enzyme - Source URL: - Dashboard article: Abstract: Antimicrobial resistance (AMR) is a global threat to human health and prosperity. Antimicrobial resistance breakers (ARBs), molecules that reverse AMR e.g., by blocking enzymes involved in bacterial membrane remodelling, are an attractive solution to restore the efficacy of existing antibiotics. We have identified several hits from in silico screening that restore polymyxin B (PmB) efficacy at concentrations of 250-500 µM in the clinically relevant ESKAPE pathogen Enterobacter bugandensis. The compounds act in tandem with PmB, but not with other antibiotics. They have no significant antibacterial effect themselves and may therefore be regarded as true resistance breakers. Preliminary results suggest that at least two of the three hit compounds may act via covalent bond formation with a target enzyme. ## BDE-47 and male infertility: integrated network toxicology, single-cell transcriptomics, and in vitro evidence of GSTM1 downregulation and ferroptosis-related molecular remodeling in Leydig cells - Source: Frontiers in Immunology (journals) - Date: 2026-09-21T00:00:00Z - Categories: Docking & Screening - Authors: Xin-Yao Zhu, Shuang Wang, Tao Li, Yu-Qi Li, Tao Zhou - Journal: Frontiers in Immunology - DOI: 10.3389/fimmu.2026.1956997 - External ID: 32e23a1cdf6609a474962984193f2d56b9a1576d - Keywords: molecular docking, molecular features, molecular dynamics - Source URL: - Dashboard article: Abstract: BDE-47 is a widely detected polybrominated diphenyl ether with potential male reproductive toxicity, but the key testicular cell populations and molecular basis remain unclear. We integrated network toxicology, single-cell transcriptomic data from BDE-47-exposed mouse testes and human idiopathic non-obstructive azoospermia (NOA), and bulk transcriptomic data. High-dimensional weighted gene co-expression network analysis, exploratory metacell-based feature prioritization with SHapley Additive exPlanations analysis, ferroptosis-driver gene signature scoring, exploratory immune deconvolution, molecular docking, 100-ns molecular dynamics simulation, and TM3 cell experiments were performed. We identified 113 shared candidate genes derived from BDE-47-related and NOA-associated gene sets, mainly enriched in xenobiotic metabolism, glutathione metabolism, oxidative stress, ferroptosis, and reproductive endocrine pathways. Mouse and human single-cell analyses localized the computational candidate signature to Leydig cells as one of the enriched somatic cell populations. GSTM1 was selected as an exploratory candidate based on multialgorithm screening, relative SHAP contribution, expression consistency across evidence layers, and redox-detoxification biological plausibility; it was expressed at lower levels in human NOA Leydig cells and bulk testicular tissue. At the cell level, lower GSTM1 expression was associated with a higher ferroptosis-driver gene signature score. Molecular simulations suggested potential structural compatibility between BDE-47 and GSTM1. In TM3 cells, BDE-47 reduced cellular metabolic activity and GSTM1 protein expression, decreased Gpx4 and Slc7a11 mRNA expression, and increased Acsl4 mRNA expression; ferrostatin-1 partially attenuated these transcriptional changes. BDE-47-related molecular evidence and human NOA-associated evidence showed molecular convergence around GSTM1 and ferroptosis-related molecular features in a Leydig-cell context. In TM3 cells, BDE-47 exposure was accompanied by GSTM1 downregulation and ferroptosis-related transcriptional remodeling. These observations are hypothesis-generating and do not establish direct binding, GSTM1-mediated effects, ferroptotic cell death, or a causal relationship between BDE-47 exposure and human NOA. ## Collaborative Laboratory Edge Networks - Source: ChemRxiv (preprints) - Date: 2026-09-21T00:00:00Z - Categories: Reaction Informatics - Authors: Andre Williams, Shanice Brown, Kevin Thompson, Jamar Campbell, Alicia Morgan - DOI: 10.26434/chemrxiv.15009129/v1 - External ID: 10.26434/chemrxiv.15009129/v1 - Keywords: retrosynthetic - Source URL: - Dashboard article: Abstract: Chemistry language models now predict reactions and retrosynthetic routes with useful accuracy, but they assume that proprietary reaction records can be pooled into a central cluster and that inference runs on abundant accelerators. Neither assumption holds in pharmaceutical and fine-chemical practice, where yield records cannot leave the plant network, the points of use are heterogeneous resource-limited sites, and different laboratories occupy disjoint regions of chemical space. We present ChemFedFusion, a heterogeneity-aware framework for federated chemistry-model training and offloading over collaborative laboratory edge networks. ChemUCP is a chemistry-aware utility–cost predictor that scores a task–node pair by expected chemistry gain, expected latency, and an explicit out-of-distribution risk of emitting unparsable or unreliable chemistry, using a dual-channel tokenizer and a scaffold-coverage encoder that is corrected online from execution feedback. CSAW replaces sample-count aggregation with the product of a scaffold-space complementarity term, a site reliability term, and a weakened volume guard, amplifying rare chemistry while suppressing noisy sites. ARCO couples offloading and aggregation through a shared adaptation matrix and an event-driven asynchronous aggregator with a staleness bound, so that successful offloads teach the aggregator which node is competent for which chemistry task. On Mol-Instruction and USPTO-50K, ChemFedFusion reaches 0.883 Exact and 51.36% Top-1, and on the FedChem suite it reduces regression RMSE by up to 9.99% and classification ROC-AUC improvements are consistent across all nine datasets and three heterogeneity levels. The framework reduces convergence rounds by 22.4% and raises the fraction of answers that are both on time and chemically valid by 8.8 percentage points. A failure analysis shows that the remaining errors are dominated by synthetically infeasible routes that parse correctly, which no validity-based router can detect. ## Computational de novo discovery of novel mTOR inhibitors using recurrent neural networks and integrated in silico approaches including docking and DFT - Source: Discover Chemistry (journals) - Date: 2026-09-21T00:00:00Z - Categories: Docking & Screening, ADMET & Safety, Design de novo - Authors: Amisha Bisht, Sanjay Kumar, Subhash Chandra - Journal: Discover Chemistry - DOI: 10.1007/s44371-026-00994-x - External ID: da6e4c8de9ab0a602434fd3f1c7350eebe71e614 - Keywords: de novo drug design, molecular docking, ChEMBL, drug likeness, toxicity prediction, recurrent neural, RNN, ADMET, pharmacokinetic, kinase, DFT, Density Functional Theory - Source URL: - Dashboard article: Abstract: The mechanistic target of rapamycin (mTOR) is a key serine/threonine kinase that regulates cell growth, metabolism, and survival, making it an important therapeutic target for cancer and other diseases. In this study, a Recurrent Neural Network (RNN)-based de novo drug design approach was employed to discover novel mTOR inhibitors. The model was fine-tuned on 5,178 experimentally validated inhibitors from the ChEMBL database (CHEMBL2842), generating 200 new candidate molecules. The candidates were evaluated using a comprehensive in silico pipeline including molecular docking, drug-likeness, ADMET profiling, toxicity prediction, synthetic accessibility, Tanimoto similarity, and Density Functional Theory (DFT) calculations. Molecular docking identified four promising leads (S000031, S000046, S000090, and S000043) with strong binding affinities ranging from − 12.2 to − 10.3 kcal/mol, outperforming or matching the reference inhibitor Torin2. These compounds showed stable interactions with key active-site residues and favorable pharmacokinetic profiles. DFT analysis revealed desirable electronic properties, including smaller HOMO-LUMO gaps, higher electronegativity, and increased electrophilicity. Among them, S000031 emerged as the most promising candidate due to its superior binding affinity, structural novelty, and electronic characteristics. This integrated computational framework demonstrates an effective strategy for identifying potential mTOR inhibitors, though experimental validation is essential to confirm their therapeutic efficacy. ## Constrained Geometry Catalysts in Ethylene Polymerization: From Mechanistic Understanding to High-Throughput Virtual Screening - Source: Organometallics (journals) - Date: 2026-09-21T00:00:00Z - Categories: Property Prediction, Library Design - Authors: Shi-Yun Feng, Rui-Hong He, Xu-Bin Wang, Qi Yang, Jing-Shuang Dang - Journal: Organometallics - DOI: 10.1021/acs.organomet.6c00254 - External ID: f9183b3e20aad74282ae5f2005e0e29d899004cb - Keywords: Virtual Screening, Gradient Boosting, density functional theory, DFT - Source URL: - Dashboard article: Abstract: The rational design of constrained geometry catalysts (CGCs) for olefin polymerization remains challenging due to the intricate interplay between steric and electronic effects and the prohibitive computational cost of exploring large chemical spaces using quantum-chemical methods. In this work, we develop a mechanistically informed machine learning framework that bridges semiempirical GFN2-xTB descriptors with density functional theory (DFT)-computed activation barriers for ethylene insertion. DFT calculations identify the first insertion as the rate-determining step (RDS), enabling construction of a physically motivated surrogate model based on the cationic methyl active species. Benchmarking across eight regression algorithms identifies Gradient Boosting Regressor as the optimal predictor, delivering DFT-comparable predictive accuracy while reducing computational cost by several orders of magnitude. SHapley Additive exPlanations analysis reveals that metal electrophilicity is the primary factor governing the activation barrier of the RDS, providing direct mechanistic interpretation of the learned structure–activity relationships. Application of the model to a virtual library of approximately 1.4 × 105 chemically valid CGCs via a high-throughput virtual screening workflow rapidly identifies low-barrier candidates and uncovers robust design principles, particularly the beneficial role of electron-donating cyclopentadienyl substituents. The present framework establishes an interpretable and efficient paradigm for accelerated polyolefin catalyst discovery. ## Cross-molecular active learning for the discovery of antimicrobial polyacrylamides - Source: Matter (journals) - Date: 2026-09-21T00:00:00+00:00 - Authors: Shoshana C. Williams, Anna Makar-Limanov, Gabriel Greenstein, Xinyu Liu, Alessio Fragasso, Priya Ganesh, Noah Eckman, Alexander N. Prossnitz, Changxin Dong, Christine Jacobs-Wagner, Lynette Cegelski, Hector Lopez Hernandez, Eric A. Appel - Journal: Matter - Source URL: - Dashboard article: Abstract: A cross-molecular machine learning pipeline was developed by using abundant antimicrobial peptide data for training before making predictions on the antimicrobial activity of synthetic copolymers. The predicted copolymers were synthesized and characterized, demonstrating favorable safety profiles and antimicrobial efficacy. They were shown to work through a membrane-disruption mechanism that helps overcome traditional resistance mechanisms. Further, they exhibit efficacy against difficult-to-target biofilm-associated bacteria. ## Cryptotanshinone and 15,16-dihydrotanshinone I from Salvia miltiorrhiza: integrated in vitro and computational evaluation of their antidiabetic potential - Source: RSC Advances (journals) - Date: 2026-09-21T00:00:00Z - Categories: Docking & Screening, ADMET & Safety, Free Energy & MD - Authors: M. C. Nguyen, Minh-Tri Le, Huynh Nguyen Khanh Tran - Journal: RSC Advances - DOI: 10.1039/d6ra06052h - External ID: 17bca5aa0f78a0d3ef485f385fdc1686831f8eff - Keywords: Lipinski, molecular docking, molecular dynamics, CYP inhibition, ADMET, pharmacokinetic, IC50 - Source URL: - Dashboard article: Abstract: Salvia miltiorrhiza Bunge (Lamiaceae) is a traditional medicinal plant rich in bioactive tanshinones with diverse pharmacological properties. However, comparative information regarding experimentally measured activity and computational interaction profiles of individual tanshinones remains limited. This study aimed to compare cryptotanshinone (1) and 15,16-dihydrotanshinone I (2), isolated from S. miltiorrhiza, using an integrated in vitro and computational approach. The compounds were obtained through activity-informed phytochemical fractionation and evaluated for inhibition of α-glucosidase at 3, 10, and 30 µM. Concentration-response data were analyzed by nonlinear regression. Their computational interaction profiles were further investigated by molecular docking against selected diabetes-related proteins, followed by molecular dynamics simulations, MM/GBSA calculations, and in silico ADMET/toxicity predictions. Compound 2 showed stronger yeast α-glucosidase inhibition than compound 1, with IC50 values of 7.76 µM (95% CI: 7.43–8.10 µM) and 10.64 µM (95% CI: 9.62–11.77 µM), respectively (p < 0.0001). In contrast, the docking preferences were target dependent: compound 1 produced more negative docking scores for human MGAM-C, PTP1B, and glucokinase, whereas compound 2 ranked more favorably for GSK-3β, NLRP3, PPAR-γ, and GLP-1R. This lack of concordance between yeast α-glucosidase inhibition and human MGAM-C docking indicates that docking rankings should not be interpreted as direct evidence of target inhibition. Molecular dynamics simulations revealed target- and ligand-dependent binding behavior, while MM/GBSA calculations yielded negative estimated binding free energies for the investigated complexes. Although both compounds satisfied Lipinski's rule, the predicted low gastrointestinal absorption, extensive plasma protein binding, CYP inhibition, and toxicity alerts indicated important developability limitations. Overall, compound 2 exhibited the stronger experimentally measured yeast α-glucosidase inhibition, whereas the computational findings provide target-specific hypotheses rather than confirming multitarget antidiabetic activity. Further studies using purified human enzymes, mechanistic assays, experimental pharmacokinetic and toxicological evaluations, and in vivo models are required to establish their biological and therapeutic relevance. ## Data-Driven Performance Prediction of Polymer Solar Cells Using Multi-Layer Perceptron with Bayesian Hyperparameter Tuning - Source: Materials Research Express (journals) - Date: 2026-09-21T00:00:00Z - Authors: Sun Ping - Journal: Materials Research Express - DOI: 10.1088/2053-1591/aeaacc - External ID: 0d396e4ed16d7b5912e1249bd6eb8d52244bf965 - Keywords: inverse design, random forest, CNN - Source URL: - Dashboard article: Abstract: Polymer solar cells (PSCs) rely on labor-intensive trial-and-error experiments to optimize power conversion efficiency (PCE), due to complex coupling of molecular electronic properties, blend morphology and fabrication parameters. This work develops an interpretable prediction framework integrating multi-layer perceptron (MLP) and Bayesian optimization (BO) to map multi-dimensional material descriptors to device PCE. A dataset of 1842 PSC records with 48 material and process features is preprocessed via missing value imputation, standardization and one-hot encoding. Dropout and L2 regularization are applied to suppress overfitting, and Bayesian optimization auto-tunes model hyperparameters within 50 trials. Comparative tests against SVR, random forest and shallow CNN show the MLP-BO model achieves a test R² of 0.962 and MAPE below 3%, outperforming all baselines. Permutation feature importance identifies optical bandgap, donor-acceptor ratio and LUMO offset as dominant efficiency factors, consistent with photovoltaic fundamentals. The framework retains high accuracy under variable light and temperature, and supports inverse design to screen optimal blend systems for target PCE, offering an efficient tool for rapid organic photovoltaic material discovery. ## De novo design of modular insecticidal scaffold enhances cotton resistance to pink bollworm and whitefly - Source: Journal of Cotton Research (journals) - Date: 2026-09-21T00:00:00Z - Categories: Docking & Screening, Design de novo - Authors: Saira Azam, Aftab Ahmad, Ayesha Latif, Naila Shahid, A. Yasmeen, A. Imran, R. Mammadova, Muhammad Awais, Samina Hassan, A. Shahid, T. Husnain, Abdul Qayyum Rao - Journal: Journal of Cotton Research - DOI: 10.1186/s42397-026-00283-z - External ID: ed31c8afb61e86342e6e328c20d2724eef942878 - Keywords: De novo design, Molecular docking, receptor - Source URL: - Dashboard article: Abstract: A variety of insect pests pose a serious threat to global cotton production. Among these, pink bollworm (PBW) and whiteflies are the most devastating, resulting in substantial yield losses and increased pesticide use. Conventional insect-resistance strategies, which primarily rely on single Bacillus thuringiensis (Bt) genes, often face limitations due to the development of resistance in target pests. To overcome these challenges, the fusion of multiple insecticidal proteins, including Bt toxins and plant-derived lectins, offers a synergistic approach to broaden the spectrum and durability of resistance. The modular insecticidal scaffold exhibited a distinct promoter-dependent expression pattern, highlighting its potential as a customisable platform for targeted cotton protection. Molecular docking revealed favorable interactions of Cry1Ac, Cry1Ab, Vip3A with the pink bollworm cadherin receptor, whereas ASAL showed favorable binding to mannose residues, representing its recognition of mannose-containing glycoconjugates on midgut cells, supporting the structural compatibility of the de novo assembled scaffold. The scaffold carrying these domains was synthesised and cloned into the pCAMBIA1302 vector under constitutive and fibre-specific promoters, and subsequently transformed into a local cotton variety. Transgenic lines harbouring the fibre-specific promoter (GhSCFP) displayed significantly elevated expression in bolls, reaching over a 100-fold increase compared with leaves, whereas the constitutive promoter (CaMV35S) drove higher expression in foliage, consistent with its broader activity profile. Promoter-guided expression patterns strongly influenced insecticidal outcomes: fibre-specific lines showed substantial mortality in PBW (up to 90–95%), whereas constitutive promoter-driven lines exhibited greater suppression of whiteflies (approximately 60–70%). The observed correlation between spatial expression and pest-specific lethality indicates that the scaffold functions as a cohesive insecticidal unit, effectively integrating multiple toxin domains while maintaining spatial regulation. This dual strategy of boll-targeted PBW protection and whole-plant whitefly resistance represents a significant advancement over single-gene Bt approaches, providing a more resilient and comprehensive resistance framework for cotton improvement. Our findings indicate that promoter selection is a crucial factor influencing tissue-specific expression and insecticidal efficacy. The synergistic combination of fibre-specific and constitutive promoters establishes a dual framework for targeted boll protection and broader pest management. This modular, promoter-driven approach holds considerable promise for the development of next-generation, sustainable pest-resistant cotton suitable for integrated pest management strategies. ## Deep Learning-Guided Interface Engineering Stabilizes Oligomeric Enzymes - Source: ACS Catalysis (journals) - Date: 2026-09-21T00:00:00+00:00 - Authors: Wei-Jie Zhan, Yang Zhuo, Zhi-Hao He, Xin-Yi Lu, Zhi-Jun Zhang, Xiao-Yu You, Qing-Chao Jiang, Kun Shi, Hui-Lei Yu - Journal: ACS Catalysis - DOI: 10.1021/acscatal.6c05133 - Keywords: Molecular dynamics - Source URL: - Dashboard article: Abstract: The thermostability of oligomeric enzymes is often limited by the intrinsic flexibility of subunit interfaces. Herein, we report a deep learning-driven interface engineering strategy (DeepIE) to systematically stabilize oligomeric enzymes. Applied to a dimeric formate dehydrogenase, we de novo redesigned six flexible regions at the dimer interface. The lead variant retains wild-type catalytic activity while exhibiting an ∼500-fold increase in half-life at 50 °C. Molecular dynamics analyses revealed that reduced local flexibility and strengthened interfacial hydrophobic packing underpin the enhanced thermostability. This work establishes an artificial intelligence-driven, generalizable framework for rational thermostabilization of oligomeric biocatalysts, effectively overcoming the activity–stability trade-off. ## DeepGVS: a bimodal deep learning framework integrating coding-sequence and protein-structural representations for virulence factor prediction - Source: Bioinformatics (journals) - Date: 2026-09-21T00:00:00+00:00 - Authors: Yan Miao, Tingting Zou, Zhenyuan Sun, Yuming Zhao, Guohua Wang - Journal: Bioinformatics - DOI: 10.1093/bioinformatics/btag698 - Source URL: - Dashboard article: - Code: Abstract: Motivation Virulence factors (VFs) mediate host adhesion, invasion, immune evasion and toxin-mediated damage, making accurate VF prediction important for understanding bacterial pathogenesis and antimicrobial intervention. Existing predictors mainly use one-dimensional (1D) protein sequences, overlooking complementary coding DNA sequence (CDS)-level information and three-dimensional (3D) structural topology. Results We propose DeepGVS, a bimodal deep learning framework integrating CDS-derived features with protein sequence and structural representations for VF prediction. DeepGVS extracts multi-scale sequence-composition features from CDS. Concurrently, it employs a parallel graph attention network (GAT) and bidirectional Mamba (Bi-Mamba) architecture to process ESMFold-predicted structures and residue representations, capturing spatial and long-range dependencies. A neural additive model (NAM) functions as a meta-learner to integrate base-classifier predictions. DeepGVS was evaluated on the unchanged accession-level independent test set of Dataset\_B and achieved an accuracy of 87.50%, corresponding to absolute improvements of 6.30, 2.60 and 1.40 percentage points over the published benchmark values of DeepVF, GTAE-VF and PLMVF, respectively. The incremental benefit of bimodal integration was dataset- and metric-dependent, and additional taxonomic analyses identified taxonomy as a potential confounding factor that does not fully reproduce the performance of the complete model. Availability and implementation Source code, datasets and pretrained models are available at https://github.com/guoguo26/DeepGVS. The archived code version used for the reported experiments is available at https://doi.org/10.5281/zenodo.21756890. Supplementary information Supplementary data are available online. ## Design, Synthesis, and In Vitro Evaluation of Pinostrobin-Based Hydrazones as Potent α-Glucosidase Inhibitors - Source: ACS Omega (journals) - Date: 2026-09-21T00:00:00+00:00 - Categories: Property Prediction, Docking & Screening, Free Energy & MD - Authors: The Thanh Ngo, Borwornlak Toopradab, Kowit Hengphasatporn, Phornphimon Maitarad, Yasuteru Shigeta, Thanyada Rungrotmongkol, Warinthorn Chavasiri - Journal: ACS Omega - DOI: 10.1021/acsomega.6c03489 - Keywords: QSAR, Molecular docking, molecular dynamics, dissociation constant, Enzyme, IC50, binding free energy - Source URL: - Dashboard article: Abstract: Twenty-four pinostrobin hydrazone derivatives (TN01–TN24) were synthesized through two continuous steps from pinostrobin, hydrazone, and aldehyde derivatives, well characterized and screened for inhibitory activity against yeast α-glucosidase (Saccharomyces cerevisiae). To further expand the chemical space, a QSAR-guided approach was employed, leading to the identification of four additional candidates (TN25–TN28). Eleven compounds exhibited >80% inhibition at 50 μM, outperforming acarbose, with most displaying IC50 values below 10 μM. Among them, TN27 showed the highest potency with an IC50 of 0.85 μM. Enzyme kinetic analysis confirmed a competitive inhibition mechanism, yielding a dissociation constant (Ki) of 3.56 μM. Molecular docking suggested favorable binding of TN27 within the active site of yeast α-glucosidase from Saccharomyces cerevisiae. To elucidate its binding mechanism, 500 ns molecular dynamics simulations were conducted in three independent replicas. Analysis of converged trajectories enabled end-point binding free energy calculations using MM/GBSA and QM-MM/GBSA methods, along with per-residue energy decomposition. These analyses revealed stable binding patterns driven by hydrophobic interactions and persistent hydrogen bonding with key catalytic residues. Collectively, the integrated experimental and multiscale computational results establish TN27 as a potent α-glucosidase inhibitor and demonstrate an effective strategy for rational lead identification and mechanistic validation. ## Development of an Olfactory Receptor (OR5K1)-Based Biosensor for Pyrazine Detection in Foods - Source: Foods (journals) - Date: 2026-09-21T00:00:00Z - Authors: Le-Le Zhang, Yan-Ping Chen, Suet Yee Tan, Zhi-Ping Xie, Xi-Chang Wang, Yuan Liu - Journal: Foods - DOI: 10.3390/foods15183355 - External ID: cb355b55ee908e28bd344776db89bd5f78ae7bf9 - Keywords: molecular docking, Receptor - Source URL: - Dashboard article: Abstract: Pyrazines are the key odorants in food contributing to baked and nutty aromas. In order to improve practicality, a novel electrochemical olfactory biosensor for pyrazine analysis was developed using an olfactory receptor (OR5K1) as the recognition element and AuNPs-PB/ZIF-8@SWCNT/Ti3C2 MXene as the sensing matrix. The sensing application and molecular recognition mechanism of OR5K1 toward pyrazines were explored using molecular docking and in silico site-directed mutagenesis. We explored the sensing application and molecular recognition mechanism of OR5K1 toward pyrazines using molecular docking and in silico site-directed mutagenesis. The sensor achieved a linear detection range of 10−14 to 10−9 M with a low detection limit of 10−14 M. The biosensor exhibited significantly higher current responses toward pyrazine compounds compared to interfering substances, including ethanol, hexanal, acetone, and phenol, demonstrating excellent selectivity, and retained 79% of its initial signal after 12 days of storage, indicating good stability. The change in the reduction peak current (ΔI) in the presence of pyrazine was used as an analytical signal. When applied to four malt samples (pilsner, munich, crystal, and caramel malts), the biosensor showed ΔI responses ranging from 4.6 ± 0.2 µA to 108 ± 6 µA, which corresponded well with the total pyrazine contents determined by GC-TOF/MS (0.092–0.391 mg/kg), with a correlation coefficient of 0.96. Molecular docking revealed binding energies ranging from −3.9 to −6.0 kcal/mol, suggesting spontaneous interactions between OR5K1 and pyrazines, with Leu14, Met81, Asn84, Phe17, Phe85, and Lys90 identified as potential key residues and hydrogen bonds, hydrophobic interactions, and π−π stacking as primary driving forces. This work provides a sensitive and selective biosensor for pyrazine detection, and the elucidated recognition mechanism offers a molecular basis for understanding roasted aroma perception, supporting applications in food quality control and flavor analysis. Collectively, this study offers new insights for designing olfactory receptor-based electrochemical biosensors and facilitates future exploration of food aroma–receptor interaction mechanisms. ## Diabetes question answering method based on knowledge graph and large language model - Source: Discover Artificial Intelligence (journals) - Date: 2026-09-21T00:00:00Z - Categories: LLMs & Agents - Authors: Yue-Long Zhang, Song-Pu Li, Tianrui Lyu, Xiao-Sheng Yu, Ting-Yao Jiang, Xiao-Long Li, Guo-Qiu He, Wen-Yuan Zhou - Journal: Discover Artificial Intelligence - DOI: 10.1007/s44163-026-02078-2 - External ID: 81bd53b4d3907a60d4ceccdd86c190528cefbb7c - Keywords: LLM, BERT, CNN - Source URL: - Dashboard article: Abstract: Conventional medical Question Answering (QA) systems often suffer from insufficient domain-specific knowledge grounding, limited semantic understanding, and weak answer traceability. To address these issues, this study proposes a decision-support QA framework that integrates structured retrieval over a diabetes knowledge graph with retrieval-augmented generation (RAG) and large language model (LLM)-based answer synthesis. First, a diabetes knowledge graph is constructed from the DiaKG corpus with explicit entity and relation types. A Qwen2.5–1.5B model fine-tuned with LoRA is used for named entity recognition, and a BERT-CNN hybrid model is designed for intent classification. In the answer generation stage, multi-hop Cypher queries over the knowledge graph are combined with guideline-oriented semantic retrieval. The two evidence sources are then separately presented in the prompt to support traceable answer generation by GLM-4-Flash through LangChain. Experimental results on component-level tasks show that the proposed method achieves disease entity recognition F1 of 81.02%, drug entity recognition F1 of 85.27%, and intent classification accuracy of 84.31%. The current system is intended as a medical information-service and decision-support tool rather than a substitute for professional clinical consultation. Further end-to-end clinical evaluation, ablation analysis, robustness testing, and expert-rated answer assessment are still required before deployment in patient-facing scenarios. ## Discovery of Novel Dual-Functional Phenylthiazole-Benzamide Hybrids as Succinate Dehydrogenase-Inhibiting Fungicides with Promising Phloem Mobility - Source: Journal of Agricultural and Food Chemistry (journals) - Date: 2026-09-21T00:00:00Z - Categories: Docking & Screening - Authors: Guo-Qing Mao, Jin-Chao Shi, Yao Tian, Yong Hu, Zhou Wen, Ya-Xuan Xiao, Jin-Zhou Ren, Lin-Hua Yu, Yu-Yan Xiao, Jun-Kai Li, Xiang Zhu - Journal: Journal of Agricultural and Food Chemistry - DOI: 10.1021/acs.jafc.6c06230 - External ID: 1eff7a53d899ed70505363a156342cc766a14c83 - Keywords: molecular docking, molecular dynamics, EC50 - Source URL: - Dashboard article: Abstract: To discover novel phloem-mobile succinate dehydrogenase inhibitor (SDHI) fungicides, 36 phenylthiazole derivatives incorporating amide fragments were designed, synthesized, and evaluated for their antifungal activity. Compound G3 exhibited the highest in vitro activity against Sclerotinia sclerotiorum (EC50 = 0.614 μg/mL), comparable to carbendazim (EC50 = 0.633 μg/mL) and boscalid (EC50 = 0.589 μg/mL). Further investigations revealed that G3 effectively inhibited sclerotial formation and germination, and exhibited remarkable in vivo protective (90.8%) and curative effects (80.3%) at 200 μg/mL. Mechanistic studies using SDH inhibition assays, scanning electron microscopy (SEM), transmission electron microscopy (TEM), molecular docking, and molecular dynamics (MD) simulations demonstrated that G3 shared a similar mode of action with boscalid. Phloem mobility studies demonstrated favorable phloem mobility of these compounds, and G6 as a representative could be efficiently absorbed and distributed in the plants. This study provides a novel strategy for the development of phloem-mobile SDHI fungicides. ## FEM-ANN-GNN modeling of MHD Darcy–Forchheimer Boger nanofluid flow with coupled heat–mass transfer in a Y-shaped hourglass cavity - Source: International Journal of Numerical Methods for Heat & Fluid Flow (journals) - Date: 2026-09-21T00:00:00Z - Authors: Faisal, Qadeer Raza, Zainab Bibi, Wan-Tao Jia - Journal: International Journal of Numerical Methods for Heat & Fluid Flow - DOI: 10.1108/hff-06-2026-0863 - External ID: 85236facde860868da8033fe50916c0d4c06022b - Keywords: GNN, graph neural network - Source URL: - Dashboard article: Abstract: This study aims to numerically investigate steady two-dimensional MHD Darcy–Forchheimer mixed convection flow of a Boger nanofluid (Cu/H2O) with coupled heat and mass transfer inside a Y-shaped hourglass cavity containing a cylindrical obstacle, incorporating thermal radiation, Cattaneo–Christov heat flux, Soret effects and chemical reaction to capture non-Fourier conduction and thermo-solutal transport. The coupled nonlinear partial differential equations are solved using a combined FEM–ANN–GNN strategy: the finite element method (FEM) generates accurate solutions, the artificial neural network (ANN) learns nonlinear input–output relationships and the graph neural network (GNN) preserves spatial interactions and mesh connectivity. Parametric studies span solvent fraction, relaxation, Forchheimer, Darcy, Hartmann, Richardson, radiation, thermal relaxation, nanolayer thickness, nanoparticle radius, Lewis, Soret and chemical reaction parameters. Flow circulation and transport are governed by the interplay of porous resistance, magnetic effects and buoyancy. Increasing permeability and solutal buoyancy enhance fluid motion and transport, whereas inertial resistance and thermal buoyancy suppress flow. Thermal radiation and nanolayer effects improve heat transfer, while larger nanoparticle size reduces thermal efficiency. The ANN yields rapid predictions with major computational savings and the GNN attains superior accuracy by preserving mesh connectivity, agreeing closely with FEM results. This work introduces a novel FEM–ANN–GNN framework for coupled thermo-fluidic transport in MHD Darcy–Forchheimer Boger nanofluids within a complex Y-shaped hourglass geometry. By integrating graph-based spatial learning with numerical and neural approaches, it delivers high efficiency and accuracy, enabling fast parametric analysis and real-time prediction of complex thermo-fluidic systems. ## Generating protein hydrogels with customizable stress relaxation behavior via deep learning-driven entanglement design - Source: Nature Communications (journals) - Date: 2026-09-21T00:00:00+00:00 - Authors: Puqing Deng, Yutong Wu, Hong Kiu Francis Fok, Linyan Li, Wen-Bin Zhang, Fei Sun, Hanyu Gao - Journal: Nature Communications - DOI: 10.1038/s41467-026-77607-9 - Keywords: de novo design - Source URL: - Dashboard article: Abstract: Protein hydrogels are promising artificial extracellular matrices (ECMs) for 3D stem cell and organoid culture due to their favorable stress relaxation behavior (a decrease in stress in response to strain). Inter-chain entangled motifs, in which different protein chains are interlaced, represent a powerful strategy to synthesize such hydrogels. However, designing these motifs with tailored properties such as binding energy remains a major challenge due to the difficulty of simultaneously controlling these properties while ensuring entanglement. Here, we introduce TangleDiff, a deep learning framework for the de novo design of homodimeric entangled proteins with programmable features. TangleDiff generates diverse foldable entangled sequences with an in-silico success rate exceeding 70%, markedly outperforming current models (~1%). By conditioning TangleDiff on inter-chain binding energy, we generate novel protein dimers whose binding energies closely match the specified ranges, with approximately 70% of successful designs conforming to expected values. We experimentally validate TangleDiff by designing nine homodimers targeting various binding energies; seven successfully form hydrogels, with stress relaxation dynamics correlated with specified binding energies. This work establishes a general strategy for entangled protein design, opening avenues for entanglement-based biomaterial innovation. ## HD-AIP: A Heterogeneous Dual-Stream Alignment-Free Framework for Anti-Inflammatory Peptide Prediction Based on Language Models and CT-Net - Source: Bioinformatics (journals) - Date: 2026-09-21T00:00:00+00:00 - Authors: Jiangli Li, Quan Zou, Yansu Wang, Yifeng Bai, Hao Zhou, Mengting Niu - Journal: Bioinformatics - DOI: 10.1093/bioinformatics/btag697 - Keywords: Computational screening, virtual screening, Transformer, CNN - Source URL: - Dashboard article: - Code: Abstract: Motivation Anti-inflammatory peptides (AIPs) show therapeutic potential for treating chronic and autoimmune diseases. Computational screening of these peptides remains challenging because their typically short sequences limit traditional feature extraction effectiveness, while homology-based methods incur high computational costs. Results This study proposes HD-AIP, an alignment-free heterogeneous dual-stream prediction architecture. This framework extracts peptide features in parallel from both macroscopic and microscopic perspectives. Macroscopically, HD-AIP integrates global semantic features extracted by two large protein language models, ProtT5 and ESM-2 3B, with sequence-level physicochemical properties, followed by feature selection and LightGBM classification. Microscopically, an asymmetric parallel network named CT-Net utilizes BioVec embeddings and residue-level physicochemical features, using a CNN branch to capture local motifs and a Transformer branch to model long-range dependencies. The two streams are adaptively fused via a dynamic soft ensemble strategy. On an independent test set, HD-AIP outperforms baseline models across multiple metrics including accuracy, area under the receiver operating characteristic curve, and the Matthews correlation coefficient. These results indicate that HD-AIP improves AIP prediction performance without sequence alignment. This architecture serves as an effective computational tool for the high-throughput virtual screening and candidate discovery of AIPs. Availability You can find the source code and dataset needed on GitHub(https://github.com/Zerofly0/HD-AIP). The required package files have been listed in the readme. Supplementary information Supplementary data are available at Bioinformatics online. ## HDACiAP: A Curated Database and Analytical Platform for Histone Deacetylase Inhibitors - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-21T00:00:00+00:00 - Categories: Docking & Screening, ADMET & Safety - Authors: Yuhe Xiao, Jiaping Ou, Jiatong Chen, Xiaolu Zhou, Xiaoying Fu, Jiaqi Hu, Zhipeng Ye, Guangying Chen, Wenyuan Kang - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c00572 - Keywords: virtual screening, molecular docking, bioactivity - Source URL: - Dashboard article: Abstract: Histone deacetylases (HDACs) are pivotal epigenetic regulators that are aberrantly expressed in various cancers, making them prominent targets for anticancer drug development. Herein, we present HDACiAP (https://hdac.kangsgo.cn), a meticulously curated, comprehensive database and analytical platform dedicated to HDAC inhibitors. HDACiAP currently encompasses 32,721 compounds with 123,154 bioactivity records, integrating multidimensional data such as physicochemical properties, biological activities, toxicity predictions, compound selectivity annotations, computationally generated docking poses, and protein–ligand interaction profiles. The platform incorporates built-in molecular docking and integrated machine learning and deep learning-based prediction modules, enabling users to evaluate the potency of novel compounds. Additionally, HDACiAP offers an intuitive visualization interface, multimodal search capabilities, and programmatic access via RESTful APIs. In summary, by providing a curated bioactivity resource together with computational docking annotations and predictive models, HDACiAP offers a practical platform for exploring HDAC inhibitor chemical space, supporting preliminary virtual screening, and facilitating AI-assisted discovery of HDAC-targeted compounds. ## HDACiAP: A Curated Database and Analytical Platform for Histone Deacetylase Inhibitors - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-21T00:00:00Z - Categories: Docking & Screening, ADMET & Safety - Authors: Yu-He Xiao, Jia-Ping Ou, Jia-Tong Chen, Xiao-Lu Zhou, Xiao-Ying Fu, Jia-Qi Hu, Zhi-Peng Ye, Guang-Ying Chen, Wen-Yuan Kang - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c00572 - External ID: dee546c4c9574ca09e07432363b5f1f59d3103e5 - Keywords: virtual screening, molecular docking, bioactivity - Source URL: - Dashboard article: Abstract: Histone deacetylases (HDACs) are pivotal epigenetic regulators that are aberrantly expressed in various cancers, making them prominent targets for anticancer drug development. Herein, we present HDACiAP (https://hdac.kangsgo.cn), a meticulously curated, comprehensive database and analytical platform dedicated to HDAC inhibitors. HDACiAP currently encompasses 32,721 compounds with 123,154 bioactivity records, integrating multidimensional data such as physicochemical properties, biological activities, toxicity predictions, compound selectivity annotations, computationally generated docking poses, and protein–ligand interaction profiles. The platform incorporates built-in molecular docking and integrated machine learning and deep learning-based prediction modules, enabling users to evaluate the potency of novel compounds. Additionally, HDACiAP offers an intuitive visualization interface, multimodal search capabilities, and programmatic access via RESTful APIs. In summary, by providing a curated bioactivity resource together with computational docking annotations and predictive models, HDACiAP offers a practical platform for exploring HDAC inhibitor chemical space, supporting preliminary virtual screening, and facilitating AI-assisted discovery of HDAC-targeted compounds. ## Impact of vintage and winemaking practices on the volatile fingerprint of young wines from Moriles Alto, whitin Montilla-Moriles PDO - Source: OENO One (journals) - Date: 2026-09-21T00:00:00Z - Authors: José Miguel Fuentes-Espinosa, Raquel Muñoz-Castells, Juan Moreno, T. García-Martínez, J. Mauricio, Jaime Moreno-García - Journal: OENO One - DOI: 10.20870/oeno-one.2026.60.3.9886 - External ID: fe9bac00b18b703927683d4eb08ca982b780cf29 - Keywords: chemometric - Source URL: - Dashboard article: Abstract: The volatile composition of Pedro Ximénez (PX) wines from three wineries of a traditional and high-quality zone within the Montilla-Moriles Protected Designation of Origin (PDO) in Spain was quantified over two consecutive vintages to evaluate the influence of vintage and winery-specific practices on terroir differentiation. Chemical parameters and volatile compounds (major and minor) were quantified using gas-chromatographic and chemometric techniques. Two-way ANOVA revealed that vintage and winery exerted a strong effect on the content of volatile compounds, indicating the influence of abiotic factors and winemaker technological decisions. Also, the significant vintage × winery interactions confirmed the multifactorial nature of the wine’s aromatic profile. Common oenological parameters are insufficient for terroir characterisation. Nevertheless, the contents of some major volatiles, such as acetaldehyde, ethyl acetate, isobutanol, 2-methyl-1-butanol, ethyl lactate, diethyl succinate, and 2-phenylethanol, have demonstrated their ability to authenticate wines by vintage and winery. The Principal Component Analysis (PCA) carried out on the major volatile compounds showed a well-defined structuring by winery along PC1 (40.18 % total variance), with additional separation by vintage along PC2 (31.72 % variance), revealing a year-dependent differentiation. Moreover, the PCA obtained with the contents of 51 minor volatiles quantified explained 70.68 % of the total variance (PC1 = 54.03 %, PC2 = 16.65 %). PC1 was defined by negative loadings of esters (e.g., ethyl octanoate and ethyl decanoate) and wines from 2022 clustered around negative PC1 values. Positive PC1 loadings corresponded to γ-butyrolactone, 4-ethylguaiacol, long-chain alcohols, and lactones, which were more abundant in the 2021 wines. PC2 differentiated wines by winery due to contributions from furanic and carbonyl compounds (positive loadings) as well as compounds such as acetophenone, ethyl benzoate, and decanal (negative loadings). The compound-specific behaviour observed indicates that no single marker can describe the effect of each factor on the volatile composition. These results provide a detailed chemical baseline to differentiate wines from Moriles Alto and highlight the value of this molecular profiling as a tool for understanding how the studied factors modulate wine composition within terroirs from a specific traditional PDO winemaking region. ## Improving the cryo-resistance of Egyptian buffalo semen via chlorogenic acid and rutin supplementation: molecular docking, antioxidant, and anti-apoptotic pathways - Source: BMC Veterinary Research (journals) - Date: 2026-09-21T00:00:00Z - Authors: W. A. Khalil, A. A. Ismail, M. Hegazy, M. A. El-Harairy, A. A. Elkashef, Mahmoud Moussa, S. Abdelnour - Journal: BMC Veterinary Research - DOI: 10.1186/s12917-026-05896-9 - External ID: 7a9b98e04fa81c7c034a08a234ae855f2601fc7b - Keywords: molecular docking - Source URL: - Dashboard article: Abstract: This study compared the antioxidant effects of chlorogenic acid (CGA) and rutin (RU) supplementation in freezing extenders on the cryo-resistance of buffalo spermatozoa. Over eight weeks, 56 ejaculates were collected from seven fertile buffalo (Bubalus bubalis) bulls. To eliminate individual animal variability, high-quality ejaculates were pooled and extended in a Tris–egg yolk medium. The pooled semen was divided into three treatment groups: control (no additive), 0.6 mM RU, and 150 µM CGA. All groups were cryopreserved using a standard protocol and stored in liquid nitrogen for one month prior to post-thaw analysis. The findings revealed that the addition of either RU or CGA substantially enhanced post-thaw sperm viability, progressive motility, plasma membrane integrity, and acrosomal integrity compared with the control group (P < 0.05). Furthermore, both treatments improved sperm kinematics and antioxidant profiles by significantly increasing total antioxidant capacity and catalase levels (P < 0.05). This was accompanied by a significant decline in oxidative markers (MDA, ROS, and H2O2) (P < 0.05), indicating robust antioxidant activity. Supplementation with RU or CGA significantly increased sperm viability and notably reduced apoptotic percentages (P < 0.05). Furthermore, CGA treatment resulted in the lowest total bacterial and coliform counts among all experimental groups (P < 0.05). Molecular docking predictions suggest a potential interaction between CGA and RU against target proteins antioxidant status (HSP70) and mitochondrial function (mitochondrial uncoupling protein 1 and mitochondrial cytochrome c) in sperm. Overall, supplementing with rutin and chlorogenic acid offers robust protection against cryodamage. By mitigating oxidative stress and preserving membrane stability, these antioxidants substantially enhance post-thawed quality, and physiological function of buffalo spermatozoa. ## Integration of multi-omics and machine learning reveals sodium overload related molecular subtypes and biomarkers in sepsis - Source: Scientific Reports (journals) - Date: 2026-09-21T00:00:00+00:00 - Authors: Lisan Cui, Min Han, Zongshuai Wang, Dejian Zhang, Bo Wang, Leilei Zhang, Shen Li, Xue Zhang - Journal: Scientific Reports - DOI: 10.1038/s41598-026-72045-5 - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Integrative structure of norovirus NS3 suggests a role in RNA transport - Source: Nature Communications (journals) - Date: 2026-09-21T00:00:00+00:00 - Authors: Meryl Haas, Tino Hoeksma, Jake T. Mills, Cynthia Kelley, Arnab Das, Kankipati Teja Shyam, Robin Veenstra, Owen Byford, Henrietta Hóf, Coco Niemel, Tim Donselaar, Ieva Drulyte, Frank J. M. van Kuppeveld, David Dulin, Joost Snijder, Morgan R. Herod, Daniel L. Hurdiss - Journal: Nature Communications - DOI: 10.1038/s41467-026-77946-7 - Keywords: AlphaFold3 - Source URL: - Dashboard article: Abstract: Human noroviruses (HuNoVs) are the leading global cause of acute gastroenteritis, yet no vaccines or antiviral therapies are currently approved. The non-structural protein NS3 is a membrane-bound AAA+ ATPase of superfamily 3 (SF3) with multiple proposed roles in the norovirus replication cycle. However, the structure of NS3, and the mechanisms by which it contributes to genome replication and membrane remodeling, have remained unknown. We engineered a soluble, hexameric, and catalytically active form of NS3 and determined its cryo-EM structure in the presence of a nucleotide analogue at 2.9 Å resolution. The structure adopts a split lock-washer architecture characteristic of AAA+ motors that operate via a hand-over-hand translocation mechanism. Complementary biochemical, single-molecule, and virological assays support oligomerization-dependent ATPase activity, ssRNA engagement, and the functional importance of conserved structural elements. Using integrative modeling with AlphaFold3, supported by targeted mutagenesis, we generated a full-length, membrane-associated model in which NS3 forms a continuous conduit across the membrane. This model supports a role for NS3 as a candidate membrane-spanning RNA translocase that may couple ATP hydrolysis to genome movement. This structural and functional framework helps address long-standing gaps in our understanding of norovirus replication and establishes a basis for mechanistic studies and structure-guided antiviral design. ## LANGER: Lanthanide Affinity Network via Graph and Evolutionary Representations - Source: ChemRxiv (preprints) - Date: 2026-09-21T00:00:00Z - Authors: Trung Nguyen, Christopher Cotter, Seunghyun Ryu, Cong T. Trinh, Duc Duy Nguyen - DOI: 10.26434/chemrxiv.15009150/v1 - External ID: 10.26434/chemrxiv.15009150/v1 - Keywords: binding affinity - Source URL: - Dashboard article: Abstract: The transition to sustainable energy relies heavily on rare earth elements (REEs), yet their separation remains a critical industrial bottleneck due to their nearly identical physicochemical properties. The discovery of lanmodulin (LanM), a highly selective REE-binding protein, has inspired bio-based separation strategies; however, predicting binding affinity between different LanM variants and REEs remains a significant computational challenge. Here, we present LANGER, a multimodal deep learning framework for predicting REE protein binding affinity (Kd) by synchronously integrating global evolutionary context, continuous ionic chemistry, and local three-dimensional coordinate geometry. Our architecture combines ESM2 protein sequence embeddings, REE physicochemical descriptions, and geometric graph learning (GGL) features derived from topologies extracted from protein-ion complexes. Using a comprehensive dataset of 616 evaluated LanM homologs against 15 REE ions, we demonstrate the robustness of this framework across two rigorous evaluation settings. In a cluster-based validation analysis designed to test extrapolation to held-out LanM sequence clusters, LANGER achieved a root mean square error (RMSE) of 0.212 and a concordance index (CI) of 0.757. More importantly, in a rigorous evaluation protocol that completely excludes individual ions from the training phase, the fully integrated multimodal framework achieved high predictive accuracy (RMSE = 0.119, Pearson correlation coefficient r = 0.860, CI = 0.800), demonstrating its unique ability to transfer learned physicochemical trends to unseen lanthanide elements. This work establishes a scalable, computationally accurate structure recognition platform for the discovery and design of data-driven REEselective proteins. ## Large Language Models in Chemistry and Materials Science: A Comprehensive Survey - Source: ChemRxiv (preprints) - Date: 2026-09-21T00:00:00Z - Categories: LLMs & Agents - Authors: JINSHI LI, CHENGCHUN LIU, CHENGRUI WEI, YIMI WANG, BOXUAN ZHAO, ZHIYUAN YAN, WENDI CAI, WENCHAO WU, SHUANGYAN TIAN, ZIFU WANG, WEIDA WANG, SHUFEI ZHANG, JIE ZHU, FANYANG MO - DOI: 10.26434/chemrxiv.15009184/v1 - External ID: 10.26434/chemrxiv.15009184/v1 - Keywords: LLMs, LLM, property prediction, reinforcement learning, synthesis planning - Source URL: - Dashboard article: Abstract: Large language models (LLMs) are expanding the scope of artificial intelligence in chemistry and materials science by connecting scientific knowledge, heterogeneous data, computational tools, and experimental workflows. This article presents a comprehensive survey of the foundations, representations, methods, applications, and evaluation of LLM-centered scientific systems. We develop a non-exclusive representation-by-mechanism taxonomy linking text, molecular and crystal structures, spectra, images, and numerical properties with model adaptation, cross-representation grounding, external knowledge and tools, and workflow coordination. Within this framework, we synthesize prompting, domain pre-training and fine-tuning, preference alignment and verifier-driven reinforcement learning, multimodal integration, and retrieval- and agent-based approaches. Applications span literature understanding, property prediction, molecular and materials design, synthesis planning, characterization, hypothesis generation, and closed-loop discovery. By linking method choices to input information, task definitions, and validation conditions, the framework clarifies which reported results can be compared and what scientific claims they support. We organize representative datasets and benchmarks and introduce an evaluation matrix connecting task performance with information fidelity, uncertainty, chemical and physical validity, and workflow reliability. Detailed supplementary cases document the settings and evidence underlying key comparisons. Finally, we identify research priorities in integrated scientific representations, efficient adaptation, reusable knowledge and tool interfaces, and verifiable human-machine collaboration. By linking methodological developments to scientific objectives, this survey provides a conceptual map and practical guidance for developing and assessing LLM-enabled chemical and materials discovery. ## Larvicidal Activity, GC–MS Profiling and Multitarget Molecular Docking of Essential Oils from Cymbopogon citratus, Tanacetum cinerariifolium and Salvia rosmarinus Against Anopheles arabiensis - Source: Molecules (journals) - Date: 2026-09-21T00:00:00Z - Categories: Docking & Screening, Free Energy & MD - Authors: Eric Kibagendi Osoro, Bernard Mwiti Kithetu, Alex Muthengi, Njogu M. Kimani, Bulelwa Audrey Ngcangatha, Fidelis Ngugi, Mojeed Adedoyin Agoro, Nicholas Rono - Journal: Molecules - DOI: 10.3390/molecules31183346 - External ID: 8b61fdce9f8c5bc1ad640704b53718d50a38042d - Keywords: Molecular Docking - Source URL: - Dashboard article: Abstract: Malaria remains a major public health challenge, particularly in sub-Saharan Africa, necessitating the development of environmentally sustainable alternatives to conventional insecticides. This study evaluated the larvicidal efficacy of essential oils extracted from Cymbopogon citratus, Tanacetum cinerariifolium and Salvia rosmarinus against Anopheles arabiensis larvae. The phytochemical composition of the essential oils was characterized using gas chromatography–mass spectrometry (GC–MS), while their potential molecular mechanisms of action were investigated through molecular docking against selected mosquito larval target proteins. All essential oils exhibited concentration-dependent larvicidal activity, with T. cinerariifolium (LC50 = 511.0 mg/L; 95% CI: 473.3–547.1 mg/L) and C. citratus (LC50 = 602.7 mg/L; 95% CI: 559.3–648.5 mg/L) demonstrating the greatest potency. GC–MS analysis identified sesquiterpene hydrocarbons as the predominant constituents of T. cinerariifolium, whereas oxygenated monoterpenes dominated the oils of C. citratus and S. rosmarinus. Hierarchical molecular docking followed by MM-GBSA analysis against mosquito juvenile hormone-binding protein (JHBP) and sterol carrier protein-2 (SCP-2) revealed that 87.13% and 77.23% of the identified phytochemicals, respectively, exhibited favorable binding affinities toward these targets, indicating that a substantial proportion of the essential oil constituents have the potential to interact with proteins critical for mosquito development and survival. Farnesol emerged as the most promising multitarget phytochemical, displaying consistently favorable docking scores and binding free energies against both proteins, while 1,10-di-epi-cubenol and (5Z,9Z)-farnesyl acetone exhibited the strongest target-specific interactions with JHBP and SCP-2, respectively. In addition, the lead compounds satisfied Tice’s insecticide-likeness criteria, supporting their suitability as insecticidal leads. Collectively, these findings demonstrate that the larvicidal activity of the investigated essential oils is likely mediated through the synergistic action of multiple phytochemicals acting on complementary molecular targets, highlighting T. cinerariifolium and C. citratus as promising sources of eco-friendly botanical larvicides for mosquito vector control. ## Lipophilicity and Steric Constraints Govern V-Shaped Inhibition of Human and Rat Placental 3β-Hydroxysteroid Dehydrogenase by Dialkyldimethylammonium Disinfectants: Implications for Progesterone Dysregulation - Source: Environmental Science & Technology (journals) - Date: 2026-09-21T00:00:00Z - Categories: Property Prediction, Docking & Screening, ADMET & Safety - Authors: Hui-Ling Deng, Xi-Shi Chen, Rui-Juan Gao, Hao Lin, Yang Zhu, Ren-Shan Ge, Jing Cheng - Journal: Environmental Science & Technology - DOI: 10.1021/acs.est.6c02784 - External ID: a458d87f1c4549a139ea805031ec570cf349224b - Keywords: QSAR, molecular docking, chemicals, IC50, enzyme - Source URL: - Dashboard article: Abstract: Dialkyldimethylammonium chemicals (DDAs) are quaternary ammonium disinfectants. However, their effects on progesterone synthesis in placentas remain unclear. This study aims to study structure–activity relationship (SAR) and mechanism of DDAs (C2–C18) as inhibitors of human (3β-HSD1) and rat (3β-HSD4) placental 3β-hydroxysteroid dehydrogenase/Δ5-Δ4 isomerase (3β-HSD), a key enzyme in progesterone synthesis. DDAs exhibited a distinct V-shaped SAR to inhibit 3β-HSD, which was dependent on alkyl chain length. The inhibitory potency peaked with the C12 analogue, demonstrating the highest potency in human (IC50 = 4.78 μM) and rat (IC50 = 2.08 μM) assays. Conversely, a significant reduction in activity was observed with the C18 analogue, as evidenced by the substantially higher IC50 values (human: 19.1 μM; rat: 18.68 μM). Kinetic analyses confirmed mixed/noncompetitive mechanisms via NAD+/steroid interface binding, supported by molecular docking and 3D-QSAR pharmacophores highlighting hydrophobic interactions. Cellular assays in JAr cells demonstrated reduced progesterone secretion. Network toxicology analysis identifies human 3β-HSD1 as a central hub within a pathogenic network for recurrent spontaneous abortion, suggesting its dysregulation to disrupted progesterone metabolism and key signaling pathways. These findings propose DDAs as novel placental 3β-HSD inhibitors, with implications for endocrine-disrupting effects. ## Moringa / Polyaniline Composites: Synthesis, Characterization, and Activity Against Drug-Resistant Bacterial and Fungal Pathogens with Molecular Docking Insights - Source: Egyptian Journal of Chemistry (journals) - Date: 2026-09-21T00:00:00Z - Categories: Docking & Screening - Authors: H. Eweis, Samar M. Mahgoub, Norham Yasser, H. A. Abd El-Salam, Amgad B. Khaliel, R. Mahmoud - Journal: Egyptian Journal of Chemistry - DOI: 10.21608/ejchem.2026.518139.13832 - External ID: 1f676d428761e792c32e3b78f7732fe0cc047fbd - Keywords: Molecular Docking - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Multimodal Representation Learning for Exploring Natural Product Chemical Space - Source: ChemRxiv (preprints) - Date: 2026-09-21T00:00:00Z - Categories: Cheminformatics, Property Prediction - Authors: Tarapong Srisongkram, Supreeya Paiboon, Huynh Anh Duy - DOI: 10.26434/chemrxiv.15009161/v1 - External ID: 10.26434/chemrxiv.15009161/v1 - Keywords: molecular fingerprints, molecular representation - Source URL: - Dashboard article: Abstract: Natural products represent a major source of therapeutic molecules, yet their structural diversity and limited biological annotations pose substantial challenges for computational exploration and discovery. Here, we develop a multimodal molecular representation framework that integrates chemical language, molecular fingerprints, physicochemical descriptors, and molecular graphs into a unified latent space for natural product discovery. Using structurally out-of-distribution molecular spaces, we show that molecular representation and sampling strategy jointly determine which regions of chemical space are explored. Multimodal pretraining captured chemically informative representations that enabled exploration of structurally unseen natural products, while comparisons with singlemodality representations showed that representation advantages depended on the discovery objective rather than being universally superior. The pretrained representation was transferable across anti-inflammatory and antioxidant activity predictions, and its integration with active learning enabled data-efficient exploration of biologically active molecules. Prospective evaluation of structurally unseen natural products further connected model-guided exploration with experimental antioxidant activity. These findings establish molecular representation and sampling as coupled determinants of chemical-space exploration and show ## Multiparadigm Benchmark of Molecular Docking: From Physics to Co-Folding and Hybrid Models - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-21T00:00:00+00:00 - Categories: Docking & Screening - Authors: Ao Xu, Jordy Homing Lam, Aiichiro Nakano, Vsevolod Katritch - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c01293 - Keywords: Molecular Docking - Source URL: - Dashboard article: Abstract: Generalizability of molecular docking predictions across novel protein targets and distinct small-molecule chemotypes is key to the successful application of deep learning co-folding approaches in drug discovery, as highlighted by recent benchmarks, including Runs-n-Poses \[Škrinjar, P.; et al.Nat. Struct. Mol. Biol.2026, 33, 782–794.\]. To compare generalizability across physics-based, hybrid rescoring models, and co-folding approaches, we systematically benchmarked several molecular docking tools representing these paradigms. Our results support previous observations that co-folding approaches show superior performance on targets resembling their training datasets. However, they deteriorate sharply to unacceptably low 20–40% success rates on novel dissimilar systems, consistent with memorization. In contrast, physics-based and hybrid methods exhibit greater out-of-distribution robustness, maintaining more than 60% success rates for docking protein–ligand complexes with minimal similarity to previously known complexes. Our results highlight the complementarity of AI-based approaches and methods based on physical sampling in all-atom models, in terms of their applicability range, and argue for the benefits of tighter integration. ## Multiparadigm Benchmark of Molecular Docking: From Physics to Co-Folding and Hybrid Models - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-21T00:00:00Z - Categories: Docking & Screening - Authors: Ao Xu, J. Lam, Aiichiro Nakano, V. Katritch - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c01293 - External ID: 3d8cdb2dffcabeb4f05af17aeda2f990ca66a508 - Keywords: Molecular Docking - Source URL: - Dashboard article: Abstract: Generalizability of molecular docking predictions across novel protein targets and distinct small-molecule chemotypes is key to the successful application of deep learning co-folding approaches in drug discovery, as highlighted by recent benchmarks, including Runs-n-Poses \[Škrinjar, P.; et al.Nat. Struct. Mol. Biol.2026, 33, 782–794.\]. To compare generalizability across physics-based, hybrid rescoring models, and co-folding approaches, we systematically benchmarked several molecular docking tools representing these paradigms. Our results support previous observations that co-folding approaches show superior performance on targets resembling their training datasets. However, they deteriorate sharply to unacceptably low 20–40% success rates on novel dissimilar systems, consistent with memorization. In contrast, physics-based and hybrid methods exhibit greater out-of-distribution robustness, maintaining more than 60% success rates for docking protein–ligand complexes with minimal similarity to previously known complexes. Our results highlight the complementarity of AI-based approaches and methods based on physical sampling in all-atom models, in terms of their applicability range, and argue for the benefits of tighter integration. ## Multiscale Interpretable Deep-Learning Framework for the Identification and Visualization of Deformation Stages in Molecular Dynamics Trajectory - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-21T00:00:00+00:00 - Categories: Cheminformatics - Authors: Zhengwu Long, Xinyu Wang, Lingyun You - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c01426 - Keywords: CNN, LSTM, molecular dynamics - Source URL: - Dashboard article: Abstract: Microscopic deformation stage recognition from molecular dynamics (MD) trajectories is crucial for understanding the evolution of material damage; however, traditional empirical analysis and black-box single deep-learning models lack both high-throughput spatiotemporal modeling and transparent physical interpretability. This work develops a multiscale interpretable deep-learning framework to automatically classify elastic, plastic, and fracture stages from uniaxial tensile MD trajectories and physically decode model decision logic. First, a 3D atomic trajectory is converted to 2D gray-scale image sequences; a CNN-LSTM hybrid architecture is built to jointly extract spatial atomic textures and long-time deformation dynamics, reaching 98.8% test accuracy and far surpassing spatial-only CNN baselines in both supervised classification and unsupervised clustering. More importantly, a hierarchical multiscale interpretability toolkit, including multilayer feature heatmaps, smoothed gradient saliency maps, gradient-weighted class activation maps, and regularized SHAP attribution, is integrated to quantify positive/negative feature contributions and localize model focus regions. The visualized attention zones perfectly match core physical fields (von Mises strain, atomic displacement, and nonaffine deformation), resolving the explainability gap of conventional MD data mining pipelines. This work establishes a generalizable, trustworthy AI paradigm that connects data-driven prediction to intrinsic microscale mechanical mechanisms for computational material research. ## Navigating continuous two-dimensional Mw–Ð space for deterministic inverse design of molecular weight distribution - Source: Matter (journals) - Date: 2026-09-21T00:00:00+00:00 - Authors: Byongkyu Lee, Jeewon Park, Geonheon Lee, Sangjin Yang, Jieun Kim, Yongjoon Cho, Changduk Yang - Journal: Matter - Keywords: inverse design - Source URL: - Dashboard article: Abstract: Molecular weight and dispersity strongly influence semiconducting-polymer performance but are difficult to tune independently through synthesis. A two-dimensional Mw–Ð design space is established by mixing three polymer batches with distinct molecular-weight distributions. Machine-learning performance mapping and inverse design then translate selected Mw–Ð coordinates into experimentally realizable mixing compositions, enabling predictive optimization of organic solar-cell performance within the batch-defined feasible region. ## Network-based discovery of resveratrol’s key regulators in lung squamous cell carcinoma - Source: Egyptian Journal of Medical Human Genetics (journals) - Date: 2026-09-21T00:00:00Z - Categories: Docking & Screening - Authors: Dini Maharani, J. García-Román, A. Hermawan - Journal: Egyptian Journal of Medical Human Genetics - DOI: 10.1186/s43042-026-00906-9 - External ID: 0e6f6763eb949d2721e4e9d162233ae1acea4b81 - Keywords: molecular docking - Source URL: - Dashboard article: Abstract: Resveratrol, a polyphenol primarily found in grapes, mulberries, and peanuts, has been studied for its anticancer properties, but the bioinformatics approach to this compound against the major subtype of non-small lung cancer type, lung squamous cell carcinoma (LUSC), had focused primarily on ferroptosis. This study evaluates the hypothesis that resveratrol exerts its effect against LUSC by simultaneously targeting multiple oncogenic pathways, thereby identifying genes of high clinical relevance, useful for future experimental validation. The approach of this study is combining data mining from several databases to obtain potential key regulators of resveratrol against LUSC. The hub genes were then analyzed for clinical relevancy and studied for binding with molecular docking. This study revealed that resveratrol affected proteins prognostically linked to poor patient survival when overexpressed, namely TP53, AKT1, IL6, and CCND1. However, only AKT1, IL6, and CCND1 showed a better binding score to resveratrol compared to the native ligand counterparts. This might demonstrate the potency of resveratrol in physically binding to the four protein targets to exert an inhibitory effect. This study suggested that resveratrol had potential targets against LUSC through cell proliferation regulation, PI3K/AKT, and inflammatory pathways. ## Odorant receptor structures predict the major female sex pheromone component in a moth - Source: BMC Biology (journals) - Date: 2026-09-21T00:00:00Z - Categories: Docking & Screening - Authors: Arthur Comte, Zhi-Qiang Wei, Hui Li, Wen-Zhe Xing, Riccardo Moracci, Jin Zhang, S. Fiorucci, Emmanuelle Jacquin-Joly - Journal: BMC Biology - DOI: 10.1186/s12915-026-02717-1 - External ID: 0106d03cc683531eca20664e3e2b044d1e4c3ff2 - Keywords: molecular docking, receptor - Source URL: - Dashboard article: Abstract: Species isolation in insects can be ensured by prezygotic barriers, such as species-specific sex pheromone blends that mediate sexual attraction and mate recognition. Yet, sex pheromone compositions of many species remain poorly characterized outside important agricultural pests, because of the necessity to access live insects for pheromone extraction, the difficulties in identifying the production period, and the precise identification of behaviorally relevant compounds in extracts. Here, we use an odorant receptor (OR)-guided approach to identify the major female sex pheromone component of the previously unexplored lily moth, Spodoptera picta. Genome analysis identified S. picta orthologs of the pheromone receptors OR5 and OR75, previously characterized in S. litura and S. littoralis. Structural modeling of OR5 and OR75 combined with molecular docking predicted conserved ligand binding to (Z,E)-9,11–14:OAc, which was confirmed by functional assays on the ORs. Gas chromatography–mass spectrometry analyses of female pheromone gland extracts identified this compound as the major pheromone component, and electrophysiological and behavioral assays further demonstrated that it elicits antennal responses and is sufficient to attract males. Together, these results establish an innovative receptor structure-based strategy for pheromone identification in non-model species, provide insights into the sexual chemical ecology of S. picta, and clarify the evolutionary pathway by which a unique class of pheromone receptors emerged by gene duplication and gained specificity. ## Overcoming Sampling Limitations Using Machine-Learned Interatomic Potentials: The Case of Water-in-Salt Electrolytes - Source: Journal of Chemical Theory and Computation (journals) - Date: 2026-09-21T00:00:00+00:00 - Authors: Luca Brugnoli, Mathieu Salanne, A. Marco Saitta, Alessandra Serva, Arthur France-Lanord - Journal: Journal of Chemical Theory and Computation - DOI: 10.1021/acs.jctc.6c00579 - Keywords: MACE, molecular dynamics, ab initio - Source URL: - Dashboard article: Abstract: Machine-learned interatomic potentials hold the promise to enable the modeling of highly concentrated liquids over meaningful time scales, far from reach for current ab initio electronic structure methods. Here we evaluate the performances of various MACE potentials in modeling a 21 m water-in-salt electrolyte based on lithium bis(trifluoromethanesulfonyl)imide. We test out-of-the-box foundation models, as well as both fine-tuning and from-scratch training strategies. Our simulations demonstrate that surrogate models allow us to overcome sampling limitations of ab initio molecular dynamics, reaching an excellent agreement with experimental observables such as the structure factor. We also demonstrate the benefit of fine-tuning a foundation model over training from scratch, in terms of data efficiency, but most importantly as a means to provide information regarding configurations hard to sample, such as short Li+–Li+ distances. Finally, we show that depending on the reference exchange-correlation functional, empirical dispersion correction schemes can be detrimental. All in all, our work shows that machine-learned interatomic potentials are a good fit for the modeling of highly concentrated electrolytes over long time scales. ## Plant-Derived Carbonic Anhydrase IV and IX Inhibitors: Predictive Pharmacodynamics and Therapeutic Potential - Source: Processes (journals) - Date: 2026-09-21T00:00:00Z - Categories: Docking & Screening - Authors: Cătălina Mareş, Andra-Maria Paun, M. Mernea, A. Mătanie, B. Cristea, I. Marinaș, S. Avram - Journal: Processes - DOI: 10.3390/pr14183016 - External ID: 092f0ccbcab14aff490ec513b341bf3aaf39a5f6 - Keywords: Drug likeness, Molecular docking, molecular dynamics, pharmacokinetic - Source URL: - Dashboard article: Abstract: Dermatological pathologies are multifactorial conditions that often show limited response to single-target therapies. Natural compounds may offer advantages through their pleiotropic effects on inflammation, oxidative stress, and tissue repair. This study employed an integrated in silico approach to evaluate selected phytoconstituents from Melaleuca alternifolia, Lavandula angustifolia, Tamarix ramosissima, and Curcuma longa. Drug-likeness, pharmacokinetic properties, dermal permeability, and toxicity were assessed using SwissADME and admetSAR 3.0. SwissTargetPrediction and the Similarity Ensemble Approach (SEA) identified carbonic anhydrases (CAs) among the predicted molecular targets, supporting the selection of CA IV and CA IX for structure-based analysis. Molecular docking identified favorable catalytic-site poses for several compounds, with curcuminoid and flavonoid scaffolds generally showing more favorable predicted affinities. Cyclocurcumin, tamarixetin, and scopoletin were subsequently investigated by molecular dynamics simulations. The simulations revealed scaffold- and isoform-dependent differences in interaction persistence: cyclocurcumin showed pronounced conformational flexibility, tamarixetin displayed more progressive displacement, and scopoletin showed the lowest persistence in the catalytic region. Overall, the results generate testable hypotheses regarding natural compound interactions with CA IV and CA IX, but experimental binding and enzymatic assays are required to establish CA modulation and therapeutic relevance. ## Preclinical identification of an oxadiazole-pyrazole lead targeting prostate cancer through integrated mechanistic evaluation - Source: Frontiers in Oncology (journals) - Date: 2026-09-21T00:00:00Z - Categories: Docking & Screening, ADMET & Safety - Authors: Vanktesh Kumar, Pankaj Wadhwa, Shubham Kumar - Journal: Frontiers in Oncology - DOI: 10.3389/fonc.2026.1953079 - External ID: 0340b781bd45a59a05148f002ff27731058ffef9 - Keywords: molecular docking, ADMET, receptor - Source URL: - Dashboard article: Abstract: Prostate cancer remains a major cause of cancer-related mortality among men, while resistance to androgen deprivation therapy emphasizes the need for new therapeutic strategies. Molecular hybridization of pharmacologically relevant heterocycles offers a rational approach for developing novel anticancer agents. Twenty oxadiazole–pyrazole hybrids (VS-01–VS-20) were designed, synthesized, and evaluated using molecular docking against the androgen receptor (PDB ID: 1Z95), followed by in vitro antiproliferative assessment against androgen-dependent LNCaP and androgen-independent PC-3 prostate cancer cells. The lead compound was further investigated for its effects on androgen receptor expression, intracellular reactive oxygen species (ROS), cell-cycle progression, and apoptosis using flow cytometry and fluorescence microscopy. ADMET properties were additionally predicted computationally. VS-04 demonstrated a favorable interaction profile within the androgen receptor ligand-binding pocket and exhibited potent antiproliferative activity, with IC 50 values of 394.95 ± 3.03 nM against PC-3 and 464.85 ± 4.29 nM against LNCaP cells. Docking–activity correlation analysis revealed a moderate and significant association between docking affinity and PC-3 antiproliferative activity, whereas no significant correlation was observed for LNCaP cells. VS-04 treatment produced a concentration-dependent reduction in androgen receptor expression and increased intracellular ROS levels. Mechanistic studies further demonstrated pronounced S-phase accumulation and induction of apoptosis, with the total apoptotic population increasing from 7.45% in control cells to 47.83% following VS-04 treatment. The integrated computational and experimental evaluation identified VS-04 as a promising oxadiazole–pyrazole lead with potent antiproliferative activity against prostate cancer cells. Its activity was associated with AR modulation, oxidative stress, cell-cycle perturbation, and apoptotic cell death, supporting further structural optimization and preclinical investigation of this scaffold for prostate cancer therapy. ## Quantifying Validation-Induced Optimism in Machine Learning for Food Spectroscopy: An Empirical Multi-Dataset Benchmark - Source: ChemRxiv (preprints) - Date: 2026-09-21T00:00:00Z - Authors: Ketney Otto - DOI: 10.26434/chemrxiv.15009168/v1 - External ID: 10.26434/chemrxiv.15009168/v1 - Keywords: chemometric, Random Forest, XGBoost - Source URL: - Dashboard article: Abstract: Background: Evaluating machine-learning regression models in food spectroscopy using random row-wise cross-validation ignores hierarchical experimental structures (replicate scans, aliquot extractions, packaging batches, instruments, and harvest campaigns). While theoretical warnings caution against sample leakage, the chemometric literature lacks a multidataset empirical quantification of validation-induced optimism, its impact on model selection, and its underlying physical mechanisms. Methods: We executed an empirical multi-dataset benchmark across four benchmark food spectroscopy collections encompassing 12,706 source spectra/records, with 6,376 spectra evaluated in primary regression workflows: Cargill corn near-infrared (NIR; n = 240 spectra across 80 physical grain samples and 3 spectrometers; targets: moisture and protein), edible vegetable oils Raman (Gilbraith , 2024; n = 215 spectra across 52 distinct reference titration determinations; target: peroxide value), mango fruit dry matter NIR (Anderson , 2020; n = 4,233 calibration spectra across 5 cultivars and n = 1,448 external harvest season spectra; target: dry matter content), and Tecator meat NIR (n = 240 spectra with 24 duplicate pairs \[48 rows participating in identical pairs\], leaving 216 unique spectral profiles; targets: fat and moisture). Six regression algorithms—partial least squares (PLS), Ridge regression, support vector regression (SVR-RBF), Random Forest, XGBoost, and k-nearest neighbours (kNN, serving as a diagnostic sentinel)—were evaluated across four hierarchical validation tiers: Level 1 (random row-wise cross-validation), Level 2 (sample-aware group cross-validation, reference-valuegrouped validation, or deduplicated validation), Level 3 (cultivar-aware validation), and Level 4 (external instrument and seasonal holdout). Model specifications were held fixed across validation tiers to isolate validation design as the sole manipulated factor, and all spectral standardisation was strictly fitted within training partitions to eliminate unsupervised leakage. Results: Transitioning from random row-wise to structured validation systematically eroded apparent predictive performance across all 36 random-to-structured model–task contrasts, yielding a median validation-induced optimism ΔR2 of 0.0170 (mean 0.3290, interquartile range \[0.0084, 0.3985\]) and a median root-mean-square error inflation ratio (RR\_RMSE = RMSE\_structured / RMSE\_random) of 1.1395 (mean 1.4277, IQR \[1.0417, 1.4374\], range \[0.9406, 5.2645\]). Degradation was especially pronounced under external domain transfer (median ΔR2 = 0.6399, median RR\_RMSE = 1.8199). In edible oils containing replicate scans sharing identical titration reference values, random cross-validation produced catastrophic performance inflation for non-linear models: kNN collapsed from an apparent R2 of 0.8865 (RMSE 5.40 meq O2/kg) to −2.1443 (RMSE 28.43 meq O2/kg; RR\_RMSE = 5.265), and Random Forest collapsed from R2 = 0.8437 to −0.5806 (RR\_RMSE = 3.180). This replicate leakage completely inverted algorithm rankings (Spearman ρ = −0.886, p = 0.019; Kendall τ = −0.733, p = 0.056). Algorithm robustness under external transfer was domain-specific: Ridge was the most resilient model under Mango seasonal transfer (R2 = 0.6690, compared to R2 ≈ 0.13–0.17 for tree ensembles), whereas uncalibrated Corn instrument transfer caused severe degradation in linear models (Ridge R2 = −30.04, PLS R2 = −37.05) due to uncorrected spectrometer baseline shifts. In contrast, non-linear models exhibited smaller relative degradation under sensor transfer. Cross-partition nearest-neighbour spectral similarity dropped by up to 0.0164 under domain transfer. Across tasks, this similarity drop showed an exploratory positive association with kNN error inflation (Pearson r = 0.739, Spearman ρ = 0.912), consistent with a spectral-proximity mechanism whereby high optical similarity enables flexible models to interpolate training points rather than learning transferable chemical relationships. Conclusions: Random row-wise validation creates severe validation-induced optimism, distorts algorithm rankings, and rewards models that memorise experimental artefacts. Reporting standards, mandatory specimen-level split manifests, and the routine inclusion of kNN as a diagnostic canary are essential to restore methodological integrity and field deployability to machine learning in food analysis. ## RaX-DT: An Open-Source Platform for Automated Molecular Docking Workflows and Reproducible Pipeline Validation - Source: ChemRxiv (preprints) - Date: 2026-09-21T00:00:00Z - Authors: Amir Hosein Ghavi, Bita Barazandeh Shirvan, Zohre Ganji, Zahra Jafari, Mojgan Nejabat, Farzin Hadizadeh, Narges Hashemi, Javad Akhondian, Farah Ashrafzadeh, Farnoosh Ebrahimzadeh, Mohammad Javad Jafari, Mehran Beiraghi Toosi - DOI: 10.26434/chemrxiv.15009171/v1 - External ID: 10.26434/chemrxiv.15009171/v1 - Keywords: Molecular Docking - Source URL: - Dashboard article: Abstract: Automated molecular docking workflows often fail not at the scoring engine but at the interfaces between ligand preparation, protein preparation, binding-site definition, and post-processing. We present RaX-DT, an open-source, Docker Compose–deployable platform that integrates these steps in a single web application and an optional conversational assistant. RaX-DT automates 3D ligand generation, tautomer/protonation-state handling, protein cleaning, binding-site definition via P2Rank, Fpocket, or user-specified coordinates, GNINA-based docking, and interactive result visualization. Because end-to-end automation can introduce silent failures, we validated the assembled workflow on 231 protein–ligand co-crystal structures from DUD-E and LIT-PCBA rather than treating the docking engine in isolation. Of these, 186 structures completed successfully; the top-ranked pose reproduced the crystallographic pose within 2.0 Å for 65.1% of completed structures, and the best of the five top-scoring poses reached 77.4%. Validation also exposed a box-placement defect that would have produced plausible-looking but invalid output, underscoring the value of pipeline-level quality control. RaX-DT is released under the MIT license and is intended to improve accessibility, reproducibility, and transparency in routine docking workflows. ## Revealing Design Rules for Bathochromic Narrowband Emission in ICz‐Based MR‐TADF Emitters via Data‐Driven Discovery - Source: Advanced Optical Materials (journals) - Date: 2026-09-21T00:00:00Z - Categories: Design de novo - Authors: Tian-Hao Tan, Xiao Chen, Ya-Xin Wang, Z. Shuai - Journal: Advanced Optical Materials - DOI: 10.1002/adom.71816 - External ID: b234c85f7ec489299ce47cb7b62209b4eb90f03d - Keywords: molecular generation - Source URL: - Dashboard article: Abstract: Indolo\[3,2,1‐ jk \]carbazole (ICz)‐based multiple resonance thermally activated delayed fluorescence (MR‐TADF) emitters exhibit promising high color purity emission. However, the lack of generalizable structural guidelines for suppressing spectral broadening associated with bathochromic shifted emission realized by conjugation extension has limited the full‐color emission from ICz emitter, particularly in the red region. In this work, we employed molecular generation to systematically explore ICz emitters, subsequently implemented interpretable machine learning (IML) to establish quantitative structure‐property relationships for wavelength and full width at half maximum (FWHM). Assisted by IML, two general structural patterns, para‐oriented N incorporation for red‐shifted emission and ring fusion across the C4‐C5 bond for spectral narrowing were identified. Electronic structure analyses revealed that these patterns enhance the non‐bonding orbital contribution of ICz core, thereby reducing the HOMO‐LUMO gap and suppressing electron‐vibronic coupling. Guided by both principles, we maintained narrowband emission while precisely tuning the emission color, dramatically increasing the success rate for target emission wavelength from 59% to 95% and reducing the proportion of broadband emission molecules by 80%. The extracted rules further guided manually crafting deep‐red emitter achieving CIE coordinates of (0.70, 0.30) with ultranarrow FWHM of 22.5 nm centered at 626 nm. ## S100A8/A9-TMEM24 axis drives podocyte injury by impairing autophagy: a novel theranostic target for early diabetic nephropathy. - Source: Pharmacological research (journals) - Date: 2026-09-21T00:00:00Z - Authors: Jia-Sen Shi, Heng Wang, Liu Xu, Yu-Lu Liu, Yu-Jie Hu, Chen Xu, Huan Li, Yun-Fei Liu, Yi-Wen Miao, Hong Jiang, Yi-Qi Liu, Lei Du, Qian Lu, Xiao-Xing Yin - Journal: Pharmacological research - DOI: 10.1016/j.phrs.2026.108472 - External ID: 676b09660fc69c3784af0d0f77ae9fe11615f817 - Keywords: virtual screening, Random Forest, binding affinity - Source URL: - Dashboard article: Abstract: OBJECTIVE To establish a comprehensive theranostic framework for early-stage diabetic kidney disease (DKD) by validating serum S100A8/A9 as a highly sensitive diagnostic biomarker and exploring therapeutic strategies targeting the S100A9/transmembrane protein 24 (TMEM24) signaling axis to restore podocyte autophagy. METHODS A clinical cohort and machine learning algorithms (Random Forest/Neural Networks) were employed to evaluate the diagnostic performance of S100A8/A9. In vivo (db/db mice) and in vitro (HG-stimulated podocytes) models were used to investigate the molecular mechanisms. Co-IP-LC/MS and protein truncation assays identified S100A9-interacting partners. Structure-based virtual screening was utilized to identify Shikonin as a novel TMEM24-targeting ligand, and its renoprotective effects were rigorously validated. RESULTS Serum S100A8/A9 levels were significantly elevated in early DKD, outperforming conventional markers in machine learning models (AUC up to 0.978). Mechanistically, S100A9 interacts with Transmembrane Protein 24 (TMEM24) through its SMP and C2 domains, hyperactivating the PI3K/AKT/mTOR pathway and impairing podocyte autophagy. Shikonin, identified through virtual screening, demonstrated a strong binding affinity for TMEM24. Administration of Shikonin dose-dependently inhibited the TMEM24-mediated PI3K/AKT/mTORC1 signaling, thereby restoring autophagic flux and ameliorating podocyte injury, glomerular basement membrane thickening, and renal fibrosis. CONCLUSIONS This study defines S100A8/A9 as a pivotal biomarker and therapeutic target, offering a theranostic strategy where Shikonin serves as a potent agent to ameliorate DKD progression by targeting the TMEM24-autophagy axis. ## The scorpion peptide Eval418 inhibits duck Tembusu virus replication by disrupting viral structure and early-stage infection - Source: Frontiers in Veterinary Science (journals) - Date: 2026-09-21T00:00:00Z - Categories: Docking & Screening - Authors: Yu-Ting Cheng, Qi Feng, Pei Zhang, Chun-Yan Ma, Wen-Jing Xie, An-Ping Wang, Zhi Wu, Wen-Feng Jia, Qing-Guo Wu, Lixia Miao, Shan-Yuan Zhu - Journal: Frontiers in Veterinary Science - DOI: 10.3389/fvets.2026.1944118 - External ID: 51e51ceeed2c6489f5c35dd6e00ddd62eecd5a03 - Keywords: molecular docking, MD simulations - Source URL: - Dashboard article: Abstract: Duck Tembusu virus (DTMUV) causes egg-drop syndrome and neurological symptoms in ducks, resulting in significant economic losses. Currently, no specific antiviral drugs are available for DTMUV infection. Scorpion-derived peptides have emerged as promising lead compounds for antiviral drug development. In this study, we investigated the antiviral activity and mechanism of the scorpion peptide Eval418 against DTMUV infection in vitro . DF-1 cells were infected with DTMUV at an MOI of 0.1 and treated with Eval418 (0–20 μM), with Ribavirin as a positive control. Antiviral activity was assessed by Western blot, RT-qPCR, and immunofluorescence assays targeting the DTMUV NS3 protein. Cytotoxicity and hemolytic activity were evaluated in DF-1, DEF, chicken and duck red blood cells the influence of duck serum on Eval418 was examined. Time-of-addition, free virion, attachment, and entry/fusion assays were performed, along with transmission electron microscopy and molecular docking/MD simulations of the NS3-Eval418 complex. Eval418 inhibited DTMUV replication in a concentration-dependent manner, achieving over 99% inhibition at 10 μM, with no significant cytotoxicity or hemolytic activity up to 120 μM in all tested cells. Mechanistic studies revealed that Eval418 acted primarily during the coaddition stage by directly inactivating free virions, blocking viral attachment, and inhibiting viral entry/fusion. TEM confirmed that Eval418 directly disrupted the morphological integrity of DTMUV virions, causing structural collapse. MD simulations showed that Eval418 binds stably to the NS3 binding pocket. Eval418 also inhibited two genetically distinct DTMUV isolates, whereas prolonged incubation with duck serum progressively reduced its antiviral activity. These findings establish Eval418 as a promising lead compound for the development of antiviral therapeutics against DTMUV infection as a model flavivirus. Its potential efficacy against other flaviviruses requires further investigation. ## Transferable Graph Neural Network Surrogates for Molecular Dynamics Across Crystal Symmetries - Source: ChemRxiv (preprints) - Date: 2026-09-21T00:00:00Z - Authors: Judah Immanuel, Avik Mahata, Aniruddha Maiti - DOI: 10.26434/chemrxiv.15009166/v1 - External ID: 10.26434/chemrxiv.15009166/v1 - Keywords: Graph Neural Network, GNN, Molecular Dynamics - Source URL: - Dashboard article: Abstract: We present a transferable graph neural network (GNN) surrogate framework for molecular dynamics (MD) that directly predicts atomic displacements and propagates atomistic configurations without explicit force evaluation or numerical time integration. The central objective of this work is to establish whether a common GNN formulation can represent atomic dynamics across materials with fundamentally different crystal symmetries and coordination environments. The same network architecture, feature representation, graph construction, and training protocol are applied without symmetry-specific modification to face-centered cubic (FCC) aluminum, body-centered cubic (BCC) iron, and hexagonal close-packed (HCP) magnesium. Across these distinct elemental and crystallographic systems, the framework achieves position-prediction errors on the order of 10−4 Å 2 and supports stable autoregressive propagation to nanosecond time scales. The predicted trajectories preserve thermodynamic stability, characteristic coordination-shell structure in the radial distribution functions, and temperature-dependent mean-squared-displacement behavior. In particular, the framework captures the closely spaced coordination shells of BCC iron and the anisotropic coordination environment of HCP magnesium without introducing lattice-specific representations. These results demonstrate that direct GNN-based atomic propagation can be formulated as a transferable framework across different elements, lattice symmetries, and coordination geometries, providing a pathway toward generalizable surrogate models for accelerated molecular dynamics. ## Triglyceride glucose-body mass index (TyG-BMI): mechanisms, multi-organ predictive value, and artificial intelligence-assisted risk stratification in type 2 diabetes mellitus - Source: Cardiovascular Diabetology (journals) - Date: 2026-09-21T00:00:00Z - Authors: Miao-Yan Yang, Jing Xu, Yun Sun, Ping Yang, Wei-Yan Shen, Ting Wang, La Ting, Nan Xu, Bo Cao - Journal: Cardiovascular Diabetology - DOI: 10.1186/s12933-026-03376-w - External ID: 9c2c587891e2c1d3c74c183c13f1e5cd30197a35 - Source URL: - Dashboard article: Abstract: Type 2 diabetes mellitus (T2DM) has evolved into a global public health crisis characterized by high prevalence, chronic progression, and multisystem complications, including cardiovascular, renal, hepatic, and bone metabolic disorders. Insulin resistance (IR) serves as the core pathological driver of T2DM onset and subsequent multi-organ damage. Traditional IR evaluation indicators, such as the homeostatic model assessment for insulin resistance (HOMA-IR), rely on costly insulin detection and lack suitability for large-scale grassroots screening and long-term dynamic monitoring. The triglyceride glucose-body mass index (TyG-BMI), a novel non-invasive and cost-effective composite metabolic marker calculated from routine biochemical and anthropometric parameters, has recently gained extensive attention in metabolic and diabetes research. Accumulating clinical evidence demonstrates that TyG-BMI exhibits superior performance in evaluating IR and predicting diabetic complications compared with single metabolic or anthropometric indicators. However, existing reviews mainly focus on single-organ diabetic complications, lacking a systematic integration of TyG-BMI-associated cardiorenal, hepatic, and bone biomarker abnormalities. Furthermore, the combination of TyG-BMI and artificial intelligence (AI) predictive modeling, a cutting-edge hotspot in precision diabetes management, has not been comprehensively summarized. This review systematically retrieves and analyzes high-quality literature published from 2016 to 2026, elaborates on the generation mechanism and molecular pathway basis of TyG-BMI mediating multi-organ injury in T2DM, summarizes the quantitative correlation between TyG-BMI and multisystem biomarker abnormalities and adverse clinical outcomes, compares the predictive efficacy of TyG-BMI with traditional IR markers, and innovatively discusses the application value of AI machine learning models based on TyG-BMI in diabetic risk stratification. In addition, we clarify population heterogeneity, conflicting research conclusions, and clinical application limitations of TyG-BMI, and propose future research directions focusing on unified cut-off value formulation, prospective intervention verification, and multi-modal AI model optimization. Overall, TyG-BMI is a promising integrated metabolic risk stratification tool for T2DM. The combination of TyG-BMI and AI predictive technology provides a novel strategy for early warning, hierarchical management, and precise clinical intervention of multi-organ diabetic complications, which has broad grassroots clinical translation prospects. ## Using Data Science Tools to Explore Rate Matching in a Nickel-Catalyzed Cross-Electrophile Coupling of Alkyl and Aryl Halides (Cl, Br) with a Tridentate Monoanionic Ligand - Source: Journal of the American Chemical Society (journals) - Date: 2026-09-21T00:00:00+00:00 - Categories: Reaction Informatics - Authors: Haruka Takenaka, Alexandra J. Ring, Avijit Hazra, Therese H. Wild, Samuel M. R. Powell, Tianhua Tang, Sarah E. Reisman, Matthew S. Sigman - Journal: Journal of the American Chemical Society - DOI: 10.1021/jacs.6c10810 - Keywords: cross coupling - Source URL: - Dashboard article: Abstract: Nickel-catalyzed cross-electrophile coupling (XEC) between aryl and alkyl electrophiles has emerged as an enabling synthetic method, yet predicting the outcomes of specific electrophile pairs is difficult due to the complexity of the mechanistic pathways. It has been proposed that “rate matching” between the oxidative addition of the aryl halide and the rate of formation of the alkyl radical, in addition to the stability of the Ni(II)–Ar complex, dictates the yield of the cross-coupling product. To probe this quantitatively, we developed a new tridentate monoanionic ligand that activatesalkyl and aryl chlorides and exhibits well-behaved electrochemistry, allowing the measurement of relative rate constants. This provides a framework to interrogate the rate-matching concept using both aryl and alkyl bromides and chlorides in cross-electrophile coupling. By combining an electroanalytical workflow with machine learning, the prediction of activation rate constants for all aryl/alkyl chlorides and bromides was achieved and correlated with reaction yields. We found that productive coupling occurs in a specific rate regime, and the synthetic performance could be predictably modulated by choosing the “matched” halide (either chloride or bromide) for each coupling partner. ## A discrete generative model of neuronal spiking activity on microelectrode arrays - Source: arXiv (preprints) - Date: 2026-09-20T22:31:29Z - Categories: Design de novo - Authors: Md Sayed Tanveer, Mohammed A. Mostajo-Radji, Ge Wang - External ID: 2609.23907v1 - Keywords: generative model, Generative models, autoencoder, transformer - Source URL: - Dashboard article: - PDF: Abstract: Generative models of neural activity could help characterize tissue dynamics, compare experimental conditions, and simulate population activity for applications ranging from disease and drug-response studies to closed-loop experimentation. Existing approaches, however, typically assume a fixed set of sorted neurons, whereas high-density microelectrode arrays produce extremely sparse, array-wide binary spike volumes in which the observed subset of electrodes varies across assays. We introduce a discrete generative model that represents this activity using a shared vocabulary of spatiotemporal motifs. A residual vector-quantized autoencoder learns the motif vocabulary, while a factorized masked transformer predicts where activity occurs and which motif appears at each active location. We evaluate the model on 31 assays spanning human brain organoids and acute \\emph\{ex vivo\} human hippocampal tissue. The learned motifs are broadly reused: assay identity explains only $9%$ of the entropy in motif use, and motif overlap across tissue types is comparable to overlap within them. When representation quality is evaluated independently of the generative prior, our approach achieves $5.2\\times$ the voxel-level reconstruction average precision of a matched flat tokenizer. For masked completion and free generation, the full model achieves $1.4$--$2.6\\times$ the site-level average precision of the matched generative baseline and outperforms it across all four families of generation metrics. These results establish a compact, reusable representation for array-wide spiking activity without learned assay-specific parameters, providing a scalable foundation for generative modeling across diverse neural preparations. ## ChemDoodle Update - Source: Macs in Chemistry (Macinchem Blog) (feeds) - Date: 2026-09-20T19:07:29+00:00 - Categories: Blog - Keywords: ChemDoodle - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Rachel: A general-purpose language model directs and revises retrosynthetic routes - Source: arXiv (preprints) - Date: 2026-09-20T12:50:30Z - Categories: Reaction Informatics, LLMs & Agents - Authors: Qisheng Li, Shunchao Jiang, Chen Qi, Xin Su, Da Han, Guangyong Chen - External ID: 2609.25118v1 - Keywords: LLM, GPT, retrosynthetic - Source URL: - Dashboard article: - PDF: - Code: Abstract: Retrosynthetic planning advances through decisions that reshape the remaining chemical problem: a locally plausible disconnection can leave precursors whose chemoselectivity constraints complicate the rest of the route. Existing planners often channel model proposals through search or template procedures, leaving open whether a general-purpose large language model (LLM) can itself sustain and revise route strategy. We developed Rachel, a stateful environment that executes and checks LLM-directed chemistry but prescribes neither a search policy nor a stopping rule. Without supplied reference routes or route-level solutions, GPT-5.5 achieved strict closure for 111 of 120 PaRoutes120 targets and 24 of 25 targets in the separate RF25 difficult-target cohort. RF25 was drawn largely from studies published after GPT-5.5's reported knowledge cutoff. Closure required complete routes and independent source resolution of every terminal precursor after planning. On a shared PaRoutes subset, forward-model support exceeded that of most comparator methods, and Rachel received the highest mean overall route score from both method-blinded LLM evaluators. Recorded trajectories showed continued model-proposed chemistry, with revised strategies carried into subsequent steps. Replacing LLM route decisions with fixed policies reduced strict closure to 6-15/120 despite continued local chemical execution; restricting planning support also reduced closure in RF25. Within Rachel, a general-purpose LLM coordinated successive chemical choices and revised its strategy as earlier decisions reshaped the remaining problems. ## This Week In Cheminformatics: Issue \#039 - Source: This Week in Cheminformatics (Substack) (feeds) - Date: 2026-09-20T11:41:49+00:00 - Categories: Blog - Keywords: Cheminformatics - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Physics-residual machine learning predicts oxygen-evolution catalyst activity beyond the training range from sparse polarization measurements - Source: arXiv (preprints) - Date: 2026-09-20T11:03:10Z - Categories: Library Design - Authors: Yong-Woon Kim, Jihyeok Lee, Sungtae Park, Sooseok Choi, Yung-Cheol Byun - External ID: 2609.23549v1 - Keywords: combinatorial libraries - Source URL: - Dashboard article: - PDF: Abstract: Screening oxygen-evolution catalysts on combinatorial libraries requires deciding which candidates receive the remaining measurements. The deciding activity lies beyond each candidate's measured potential window and often above every activity recorded during fitting. We predict it by physics-residual machine learning: the Tafel equation extrapolates the candidate's own measured current and slope, a learned residual attenuated with feature-space distance corrects the magnitude, and an applicability-domain score identifies predictions above the training range before measurement. In a separately fabricated 322-candidate library, 282 above the training maximum, two measurements per candidate gave a mean absolute error of 0.203 mA cm$^\{-2\}$ against 1.330 for the selected data-driven machine-learning model. Errors inside the training range remained comparable, and 35 labelled catalysts were enough to fit it. In two independent datasets the same construction lowered the overpotential error by 29 to 52%. Campaigns can therefore shorten each measurement and still rank the most active compositions. ## A substrate-first-enzymology deep-learning framework for transforming new-to-nature chemicals - Source: ChemRxiv (preprints) - Date: 2026-09-20T00:00:00Z - Authors: You-Xue Zhao, Hao-Yue Wang, Luo-Bing Meng, Yu-Xin Jia, Xi-Ruo Li, Jia-Ming Yang, Romas J. Kazlauskas, Gui-Sheng Fan, Jian-He Xu - DOI: 10.26434/chemrxiv.15008807/v2 - External ID: 10.26434/chemrxiv.15008807/v2 - Keywords: attention mechanism, chemicals, Enzyme - Source URL: - Dashboard article: Abstract: Enzyme activity data favor familiar substrates and positive outcomes, creating survivorship bias that limits predictions for new-to-nature chemistry. Here we introduce substrate-first enzymology: a chemically diverse substrate array is designed to broaden coverage of a predefined, synthetically relevant chemical space, independently of expected activity. Alcohol dehydrogenases (ADHs) are systematically assayed against this array using a unified high-throughput assay. The enzyme–substrate matrix records activity labels and turnover frequency ( k cat ) values; substrate-resolved profiles form an ADH k cat atlas. A deep learning model trained on these data classifies activity and predicts k cat for active pairs. The model outperforms existing approaches under matched training and evaluation conditions and prospectively prioritizes active substrates from underrepresented chemical space. Without binding-pocket annotations, the model’s attention mechanism highlights substrate reactive centers and catalytically relevant residues. By designing substrate coverage before measurement and retaining both positive and negative outcomes, this framework reduces historical sampling bias and supports interpretable k cat prediction beyond historically sampled chemistry. ## Antimicrobial activity of Vepris salicifolia leaf essential oil and δ-cadinene: in vitro evaluation and in silico mechanistic insights into δ-cadinene - Source: Tropical Medicine and Health (journals) - Date: 2026-09-20T00:00:00Z - Categories: Docking & Screening, ADMET & Safety - Authors: B. Paulos, Avijit Mazumder, P. Lindemann, M. Y. Yeshak, D. Bisrat, K. Asres - Journal: Tropical Medicine and Health - DOI: 10.1186/s41182-026-01063-w - External ID: 60ad51dc0c8fcaafa42b0fbd4bfbb88b3871cda3 - Keywords: drug likeness, molecular docking, ADMET, pharmacokinetic, enzyme - Source URL: - Dashboard article: Abstract: The emergence of antimicrobial resistance has intensified the search for new bioactive compounds from natural sources. This study investigated the chemical composition and antimicrobial activity of the essential oil (EO) extracted from the leaves of Vepris salicifolia (Engl.) Mziray. In addition, the molecular interactions and drug-like properties of δ-cadinene, the major constituent of the EO, were evaluated. The EO was extracted by hydrodistillation and subsequently characterized chemically using gas chromatography–mass spectrometry (GC–MS). The antibacterial and antifungal activities of the EO and δ-cadinene were assessed using disc diffusion and agar dilution assays, while molecular docking and ADMET analyses were performed to elucidate the potential mechanism of antimicrobial action and evaluate the pharmacokinetic and drug-likeness properties of δ-cadinene. GC–MS analysis identified 26 compounds, representing 84.3% of the EO. The predominant constituents were δ-cadinene (23.5%), intermedeol (18.3%), germacrene D (10.2%), α-muurolol (9.0%), and germacrene D-4-ol (7.5%). The EO and δ-cadinene effectively inhibited the growth of a wide range of Gram-positive and Gram-negative bacteria and fungi. Notably, δ-cadinene showed antibacterial activity against S. aureus ML267, with an inhibition zone of 17.8 mm, comparable to that of ciprofloxacin (18.0 mm). Furthermore, the EO exhibited greater antifungal activity against Penicillium notatum, with an inhibition zone of 14.0 mm compared with 11.0 mm for griseofulvin. Docking results showed favorable binding energies for δ-cadinene with the CrtM enzyme of S. aureus (− 7.757 kcal/mol) and the CYP51 enzyme of P. notatum (− 5.756 kcal/mol). ADMET studies further revealed that δ-cadinene has favorable drug-like characteristics, including high intestinal permeability and oral bioavailability, and did not predict any cardiotoxicity. The findings of this study indicate that V. salicifolia leaf EO, particularly δ-cadinene, represents a promising source of antimicrobial agents for future medicinal applications. Nevertheless, the predicted molecular mechanisms of δ-cadinene should be validated through enzyme inhibition assays. ## Baicalin Induces Ferroptosis in Breast Cancer Through a Novel STAT3/HO-1/GPX4 Regulatory Axis. - Source: Phytotherapy research : PTR (journals) - Date: 2026-09-20T00:00:00Z - Categories: Docking & Screening - Authors: Ting-Ting Guo, Fansu Meng, De-Tang Li, M. Mazur, Alicia m.Díaz García, M. Asim, Zhong Chen, Yi-Qi Chen, Yi-Xuan Lin, Zhenjiang Yang, Pan-Pan Wang, Yu Cai - Journal: Phytotherapy research : PTR - DOI: 10.1002/ptr.70426 - External ID: 67016a7eeaef494a79559e0a09049fb4b7fb3910 - Keywords: molecular docking - Source URL: - Dashboard article: Abstract: Breast cancer (BC) is a leading cause of cancer-related mortality worldwide. Ferroptosis is a non-apoptotic regulated cell death that has been linked to the progression of BC and represents a promising therapeutic target for this malignancy. Baicalin (BA) is recognized for its antioxidant, anticancer, and anti-inflammatory effect. Although the ferroptosis-mediated antitumor activity of BA has been widely demonstrated in various cancers, its potential to induce ferroptosis in BC remains unclear. To explore its regulation of ferroptosis in BC and identify the underlying mechanisms, we performed multi-dimensional experiments to investigate the mechanisms of BA-induced ferroptosis. Proliferatively, CCK-8 and colony formation assays were employed to measure the viability and reproductive capacity of cells, complemented by a nude mouse xenograft model for in vivo validation. Subsequently, we demonstrated BA-induced cell death via ferroptosis in vitro and in vivo in BC, accompanied by ROS and lipid peroxidation accumulation, GSH depletion, intracellular liable Fe2+ enrichment and mitochondrial damage. Notably, these effects were reversed by ferrostatin-1. The pan-caspase inhibitor Z-VAD-FMK and Fer-1 rescued BA-induced cell death, whereas necroptosis inhibitors necrosulfonamide and necroptosis inhibitor necrostatin-1 as well as autophagy inhibitor chloroquine did not exert such effects. Significantly, ferroptosis serves as the predominant cell death mechanism. Mechanistically, we performed network pharmacology analysis and molecular docking to explore the interaction between BA and STAT3/HO-1/GPX4. Functionally, BA effectively disrupted GPX4-dependent ferroptosis defense and induced BC cell death. As a natural compound modulating this ferroptosis-related axis, BA is a promising therapeutic alternative to conventional chemotherapy, which is limited by severe systemic toxicity. Collectively, this work reveals BA may regulate ferroptosis via the potential STAT3/HO-1/GPX4 signaling axis. ## Beauvericin exhibits antitumor activity in pancreatic cancer cells: Integrative analysis suggests potential modulation of the KEAP1-NRF2-HMOX1 axis and survival signaling pathways - Source: Scientific Reports (journals) - Date: 2026-09-20T00:00:00+00:00 - Categories: Docking & Screening - Authors: Xinyi Yu, Xiaodan Liu, Jie Ouyang, Liming Li, Ruoxuan Liu - Journal: Scientific Reports - DOI: 10.1038/s41598-026-72126-5 - Keywords: molecular docking, molecular dynamics, MD simulations - Source URL: - Dashboard article: Abstract: To investigate the anticancer effects of Beauvericin (BEA) and elucidate its potential mechanisms in pancreatic cancer cells, a malignancy characterized by marked heterogeneity and frequent chemoresistance. The effects of BEA on the proliferation, migration and apoptosis of Mia PaCa-2 cells were evaluated via CCK-8, colony formation, Transwell and flow cytometry assays. Systems biology was used to prioritize candidate targets, and molecular docking and molecular dynamics simulations were performed to evaluate a putative BEA–KEAP1 interaction computationally. Western blotting was used to detect the expression of key signaling pathway proteins. In Mia PaCa‑2 cells, BEA reduced cell viability (IC₅₀ = 1.45 µM), suppressed colony formation at 0.5 µM, decreased migration at 1 µM, and increased apoptosis to 50.06% at 2 µM. Docking suggested the presence of putative interactions between BEA and KEAP1 (estimated docking score/energy of -9.45 kcal/mol) and between BEA and HMOX1 (-7.82 kcal/mol). MD simulations supported the stability of the predicted BEA–KEAP1 complex (RMSD ~ 0.5 nm). BEA treatment was associated with the modulation of KEAP1/NRF2‑related signaling and reduced the phosphorylation of AKT, ERK1/2, and STAT3, which is consistent with the attenuation of key survival signaling. Proteomic profiling revealed enrichment of differentially expressed proteins involved in oxidative phosphorylation and ribosome‑related processes following BEA treatment. BEA had dose-dependent effects on Mia PaCa-2 cells and affected multiple signaling nodes. These findings support further investigations of BEA in additional PAAD models, including chemoresistant settings, to evaluate its translational potential. ## Biosynthetic Plausibility Mapper: Reaction-Aware Prioritization of Plant Metabolite Candidates Using Curated Pathway Context - Source: ChemRxiv (preprints) - Date: 2026-09-20T00:00:00Z - Categories: Cheminformatics - Authors: Lucas Jesus Patrício, Rômulo Pereira de Jesus, Ísis Tavares Vilas Boas, Fernanda Oliveira Chagas, Ricardo Moreira Borges - DOI: 10.26434/chemrxiv.15009098/v1 - External ID: 10.26434/chemrxiv.15009098/v1 - Keywords: cheminformatics - Source URL: - Dashboard article: Abstract: Untargeted plant metabolomics can generate numerous structurally plausible annotations and candidate structures, but structural similarity alone does not establish whether a compound is consistent with the biosynthetic context represented by the detected metabolome. Integrating curated biochemical relationships with chemical similarity could therefore provide an additional, interpretable layer for candidate prioritization. Here, we present Biosynthetic Plausibility Mapper (BPM), an opensource cheminformatics application that maps user-provided plant metabolite annotations onto a locally constructed PlantCyc-derived knowledge base and explicitly distinguishes detected compounds, curated pathway context, and computationally prioritized candidates. BPM integrates chemical-identity reconciliation, curated reaction and pathway relationships, structural similarity, and global-context coherence into a decomposable candidate-ranking framework. In pathway-based benchmarking against hard structural decoys, BPM v3.1 achieved AUROC and AUPRC values of 1.000 and ranked the reference compound first in all 20 trials. In a more stringent adversarial benchmark containing biosynthetically plausible alternatives, v3.1 improved AUROC from 0.920 to 0.985, AUPRC from 0.574 Biosynthetic Plausibility Mapper to 0.837, and Top-1 accuracy from 65% to 90% relative to v3. A compact phenylpropanoid–flavonoid example illustrates the separation of experimental annotations, curated context, and prioritized hypotheses, while a larger Arabidopsis occurrence dataset demonstrates application to 864 mapped compounds within a 6,714-node network. BPM therefore provides an interpretable framework for converting annotated plant metabolite lists into ranked biosynthetic hypotheses without conflating computational plausibility with compound identification. By combining curated biochemical context with structural evidence and exporting prioritized candidates for targeted interrogation of LC–MS/MS data, BPM provides a practical link between metabolite annotation, biosynthetic interpretation, and subsequent experimental investigation. ## Effects of Dietary Periplaneta americana Residue Supplementation on Porcine Ovaries and Screening Analysis of Functional Small Peptides - Source: Animals (journals) - Date: 2026-09-20T00:00:00Z - Categories: Docking & Screening - Authors: Cheng-Long Pan, You Tan, Yu-Jun Wu, Zi-Lin Li, Rong Jiang, Lin-Jie Xu, Bi-Rong Zhang, An-Ran Xu, Ran Pu, Jun-Jun Wang, Shi-Yan Sui - Journal: Animals - DOI: 10.3390/ani16182958 - External ID: 9157fd251592d8fd08c07aec729bfc0ad4404d8b - Keywords: molecular docking, molecular dynamics - Source URL: - Dashboard article: Abstract: Background: Periplaneta americana residue (PAR) has potential as a sustainable insect-derived feed ingredient. This study evaluated the effects of 3% dietary PAR on growth performance and ovarian function in finishing pigs and screened potential bioactive peptides. Methods: Thirty finishing pigs were assigned to a control diet or a diet containing 3% PAR. Growth traits, serum oxidative and inflammatory indicators, and ovarian histomorphology were assessed in vivo. Multi-omics profiling was integrated with quantitative real-time PCR, Western blotting, and immunofluorescence assays. The properties of selected peptides and their potential interactions with candidate proteins were assessed separately using in silico prediction, molecular docking, and molecular dynamics simulations. Results: PAR supplementation increased average daily gain and the ovary-to-body weight ratio and decreased serum malondialdehyde levels and follicular atresia, without significantly affecting feed intake, feed conversion ratio, or inflammatory cytokines. Multi-omics profiling associated PAR supplementation with changes in ovarian metabolism, antioxidant defense, immune regulation, and phosphorylation-related signaling. MGST2 and PPM1K were identified as candidate proteins, and their expression patterns were supported experimentally. Serum peptidomics identified GVSEIQQ as having higher abundance in the PAR group than in the control group, whereas GVEEHETL and PGIPT were detected only in the PAR group. In silico analyses predicted that these peptides were non-toxic, non-allergenic, and potentially antioxidant; these predictions require experimental validation. Molecular docking and molecular dynamics simulations suggested potential interactions of GVSEIQQ with MGST2 and GVEEHETL with PPM1K. However, these computational findings do not establish direct peptide–protein binding or functional effects in vivo. Conclusions: PAR shows potential as an alternative feed ingredient that may partially replace soybean meal while supporting growth and ovarian condition in finishing pigs. Further biochemical and in vivo studies are required to confirm the biological functions and mechanisms of the identified peptides. ## Generative AI for Drug Discovery: GPT-2 and LSTM-Based Models for Designing EGFR Inhibitors - Source: Medinformatics (journals) - Date: 2026-09-20T00:00:00Z - Categories: Cheminformatics, Docking & Screening, Design de novo - Authors: Omar Dasser, Othman Filali Benaceur, Salma Fadel, Rim Kaanane - Journal: Medinformatics - DOI: 10.47852/bonviewmedin62029449 - External ID: 161c52fc637367ea26c70fbf602be85d047c3dfb - Keywords: Lipinski, drug likeness, molecular docking, ChEMBL, generative models, LSTM, GPT, receptor - Source URL: - Dashboard article: Abstract: Effective inhibitors for the epidermal growth factor receptor (EGFR) still present a major obstacle in the development of cancer drugs. In this work, we create new EGFR inhibitors using generative artificial intelligence (AI) by fine-tuning a GPT-2 model on a dataset comprising around 500,000 molecules from the ChEMBL database. We evaluate our method against a Long Short-Term Memory (LSTM) network trained on the same dataset to create benchmarks. Before being filtered depending on Lipinski's rule of five, synthetic accessibility, and drug-likeness scores, the produced compounds were evaluated for validity, distinctiveness, and novelty. To assess their binding affinities, chosen candidates underwent molecular docking experiments against EGFR (PDB ID: 1M17). Our results show that although GPT-2 excels in producing structurally varied molecules, the LSTM model generates a larger fraction of chemically valid compounds. Many produced candidates showed good binding interactions with EGFR, therefore highlighting the possibilities of deep learning-based generative models in hastening the early phases of therapeutic development. This work presents a scalable method for creating tailored cancer treatments by showing the synergy between AI and in silico drug design. Received: 24 February 2026 | Revised: 21 July 2026 | Accepted: 3 September 2026 Conflicts of Interest The authors declare that they have no conflicts of interest to this work. Data Availability Statement The data that support this work are available upon reasonable request to the corresponding author. Author Contribution Statement Omar Dasser: Conceptualization, Methodology, Software, Validation, Formal analysis, Investigation, Resources, Data curation, Writing– original draft, Writing– review & editing, Visualization, Supervision, Project administration. Othmane Filali benaceur: Software, Visualization. Salma Fadel: Validation, Inves tigation, Writing– review & editing. Rim Kaanane: Validation, Formal analysis, Investigation, Writing– review & editing. ## Gentiopicroside Regulates EGFR and Induces Its Lysosomal Degradation to Attenuate Heart Failure. - Source: Phytotherapy research : PTR (journals) - Date: 2026-09-20T00:00:00Z - Categories: Docking & Screening - Authors: Ming-Zhen Cao, Da Ke, Jian Ni, Yuan Yuan, Zhang-Yi-Juan Nan, Xiu-Jun Dai, Heng Zhou - Journal: Phytotherapy research : PTR - DOI: 10.1002/ptr.70452 - External ID: 00ce8d22592c844cdd1f650bc2ba904c4f3ea0a4 - Keywords: molecular docking, kinase - Source URL: - Dashboard article: Abstract: Cardiac hypertrophy is a major pathological response to cardiovascular stress and a critical contributor to heart failure (HF). Gentiopicroside (GPS), a naturally occurring iridoid glycoside derived from Gentiana scabra, exhibits anti-inflammatory, antioxidant, and anti-fibrotic properties. However, its role and molecular mechanisms in cardiac hypertrophy and HF remain unclear. A pressure overload-induced cardiac remodeling model was established in C57BL/6 mice by transverse aortic constriction (TAC), followed by GPS treatment with or without the EGFR tyrosine kinase inhibitor Canertinib. Cardiac function and remodeling were evaluated by echocardiography, hemodynamic analysis, and histological staining. In vitro, neonatal rat ventricular myocytes (NRVMs) and cardiac fibroblasts (NRCFs) were used to investigate the cellular effects of GPS. EGFR knockdown, conditioned medium transfer, cellular thermal shift assay (CETSA), and lysosomal/proteasomal inhibition experiments were performed to elucidate the mechanism underlying GPS-mediated EGFR regulation. GPS markedly alleviated TAC-induced cardiac hypertrophy and fibrosis in vivo and suppressed Ang II-induced cardiomyocyte hypertrophy. GPS also reduced cardiomyocyte apoptosis and oxidative stress. Mechanistically, GPS acted as an EGFR degrader rather than a conventional kinase inhibitor, reducing total EGFR abundance without inhibiting the kinase activity of remaining receptors, thereby attenuating excessive AKT/ERK1/2 signaling. CETSA and molecular docking indicated direct GPS-EGFR interaction. GPS-induced EGFR reduction was mediated primarily through lysosomal degradation, as demonstrated by chloroquine rescue experiments, whereas proteasome inhibition had limited effects. Co-administration of Canertinib with GPS conferred no additional benefit. EGFR depletion abolished the protective effect of GPS in cardiomyocytes, indicating a cardiomyocyte-autonomous mechanism. This study identifies GPS as a regulator of EGFR stability during pathological cardiac remodeling. By promoting lysosome-dependent EGFR degradation and suppressing excessive AKT/ERK1/2 signaling, GPS attenuates cardiac hypertrophy and remodeling while preserving basal EGFR function, highlighting its potential therapeutic value for HF. ## Hydrogels at the Biology–Physics–Chemistry Interface: Design Principles Guiding Emerging Frontiers - Source: Langmuir (journals) - Date: 2026-09-20T00:00:00Z - Authors: S. Chakraborty, Biswajit Dey - Journal: Langmuir - DOI: 10.1021/acs.langmuir.6c04432 - External ID: 2aefe1bfe447c7a647a5912417f324882008188e - Source URL: - Dashboard article: Abstract: Hydrogels─three-dimensional networks retaining water contents while maintaining mechanical coherence─have become the preeminent soft material platform for applications spanning tissue engineering, drug delivery, flexible electronics, environmental remediation, and industrial manufacturing at scales exceeding thousands of metric tons annually. Yet, the transition from empirical discovery to rational design remains impeded by a fundamental characterization gap: macroscopic function emerges from molecular events at interfaces, but conventional techniques systematically fail to access these boundaries. This Review advances the work that trying to bridge this gap─from subnanometer hydrophobic driving forces to centimeter-scale tissue mechanics─constitutes the defining challenge for predictive hydrogel science. We critically examine the structural hierarchies distinguishing covalently cross-linked polymeric networks from supramolecular systems assembled through noncovalent patterns. The molecular building blocks enabling programmable assembly─peptide amphiphiles, nucleobase gelators, amino acid derivatives, carbohydrate organogelators, and metallosupramolecular architectures─are surveyed alongside design principles from molecular recognition to sample-spanning networks. The applications survey demonstrates universality across domains: biomedical systems; optoelectronic devices; catalytic platforms; nanocomposite architectures; and phase-selective systems for environmental remediation. We identify six interconnected challenges whose collective resolution will transform the field: achieving molecular-to-macroscopic prediction, integrating characterization across length scales, enabling measurement under native conditions, balancing mechanical robustness with responsive dynamics, establishing translational validation, and resolving interfacial phenomena at gel–cell boundaries. The vision for the coming decade centers on correlative, in-situ, multimodal platforms─augmented by machine learning on standardized datasets─these close the loop between molecular programming and macroscopic performance. Success will be measured not by new gelators discovered but by thecapacity to specify a desired function and design the molecular system that will achieve the transformation from hydrogel discovery to hydrogel design. ## Insecticidal activity of Metarhizium anisopliae-mediated nanoparticles and fungal metabolites against the red palm weevil, Rhynchophorus ferrugineus - Source: Scientific Reports (journals) - Date: 2026-09-20T00:00:00+00:00 - Categories: Docking & Screening - Authors: Fahad M. Alshammari, Abdulrahman S. Aldaghmi, Elsayed E. Hafez, Samia Q. Alghamdi, Laila S. Alqarni, Nada Alhathlaul, Wael Elmenofy, Maher A. Hammad, Ghada M. El-Sayed, Fatma H. Galal - Journal: Scientific Reports - DOI: 10.1038/s41598-026-69462-x - Keywords: Molecular docking - Source URL: - Dashboard article: Abstract: This study evaluated the insecticidal activity of green-synthesized metallic nanoparticles (Ag, CuO, and ZnO) and the ethanolic extract of Metarhizium anisopliae against the red palm weevil ( Rhynchophorus ferrugineus ). Physicochemical characterization confirmed the successful synthesis and stability of the nanoparticles. Among the tested materials, AgNPs exhibited the greatest larvicidal activity (LC₅₀ = 568.0 ppm), outperforming both the fungal extract and the reference insecticide abamectin under the experimental conditions. GC–MS analysis of the M. anisopliae extract identified several bioactive metabolites, including 1,2,4-benzenetriol, oleic acid, and naphthalene derivatives. Enzymatic assays demonstrated that AgNPs markedly inhibited digestive enzymes, particularly amylase and protease, indicating disruption of nutrient metabolism. Gene expression analysis revealed treatment- and time-dependent modulation of inflammatory (IL6 and TNF) and detoxification (CYP450) genes. Molecular docking further suggested that fatty acid derivatives, particularly oleic acid, exhibited favorable binding affinities toward multiple insect protein targets, supporting their potential contribution to the observed insecticidal activity. Overall, these findings provide mechanistic insights into the insecticidal effects of green-synthesized metallic nanoparticles and M. anisopliae metabolites and support their potential as promising candidates for red palm weevil management. However, further field-based studies are required to validate their practical efficacy under natural conditions. ## Interaction Mechanism of Bovine Myosin Oxidation Inhibition by Epicatechin Gallate: Insights from Molecular Docking and Thermodynamic Analysis - Source: Foods (journals) - Date: 2026-09-20T00:00:00Z - Categories: Docking & Screening - Authors: Xin Li, Mei-Juan Xu, Min Li, Xin-Ming Yin, Huang Zhang, Xue-Qin Gao, Chang-Bin Li, Zhi-Da Sun, Bi-Jun Xie, Jian Zou - Journal: Foods - DOI: 10.3390/foods15183332 - External ID: b5a96866d62a2e9182a082e01846e50fa6a9b531 - Keywords: Molecular Docking - Source URL: - Dashboard article: Abstract: Protein oxidation is a critical factor governing meat quality. Although lotus seedpod proanthocyanidins (LSPC) can inhibit protein oxidation in beef during refrigerated storage, the underlying mechanism remains unclear. This study investigated the antioxidant activity and conformation impacts of epigallocatechin gallate (ECG), the principal active constituent of LSPC, on myosin from the perspective of molecular interactions. Isothermal titration calorimetry (ITC), molecular docking, and multispectral analyses were applied to characterize ECG–myosin interactions and unravel the mechanism underlying myosin oxidation inhibition. The results showed that ECG exerted a biphasic, dose-dependent effect on myosin conformation and related functional properties. At 10–30 μmol/L, ECG significantly enhanced the radical scavenging capacity of myosin and alleviated oxidative damage to amino acid side chains. Mechanistically, ECG bound spontaneously to myosin via entropy-driven hydrogen bond and hydrophobic interactions, inducing conformational rearrangement characterized by an α-helix-to-β-sheet transition, reduced surface hydrophobicity, and stabilization of secondary and tertiary structures, thereby inhibiting oxidative denaturation of myosin. However, at 40 μmol/L, ECG still retained antioxidant capacity toward myosin, while it increased surface hydrophobicity and reduced fluorescence intensity. These changes further aggravated the structural disorder, led to protein aggregation and microstructure damage, and ultimately caused the instability of myosin. These findings underscore the necessity of optimizing ECG dosage for reliable application in meat processing and provide experimental evidence supporting ECG and LSPC as natural antioxidants for meat systems. ## Interpretable Machine Learning Frameworks for QSAR Modeling: A Comparative Study of Classical Algorithms and Transformer-CNN Architectures - Source: ACS Omega (journals) - Date: 2026-09-20T00:00:00+00:00 - Categories: Cheminformatics, Property Prediction - Authors: Fatima Ezzahra Belkhanchi, Mohamed Mbarki, Samir Chtita, Amit Kumar Halder, M. Natália D. S. Cordeiro - Journal: ACS Omega - DOI: 10.1021/acsomega.6c03768 - Keywords: QSAR, ChEMBL, CNN, Convolutional Neural, Transformer, enzyme - Source URL: - Dashboard article: Abstract: Phosphodiesterase type 5A (PDE5A) inhibitors represent an important class of bioactive molecules with therapeutic relevance across cardiovascular, neurological, and oncological disorders. In this work, we develop an interpretable Quantitative Structure–Activity Relationship (QSAR) framework that integrates classical machine learning (ML) approaches with a hybrid deep learning architecture (Transformer–Convolutional Neural Network, Transformer–CNN) to model enzyme inhibition using a curated and diverse data set from ChEMBL. Multiple structural descriptors and fingerprint-based representations were evaluated using established ML algorithms, with fingerprint-driven models, particularly those employing FCFP4 features, achieving the highest predictive performance with Matthews Correlation Coefficients up to 0.783. Comparative analysis revealed that optimized classical ML models slightly outperform the Transformer–CNN approach for this data set. Model interpretability was explored through Shapley Additive Explanations (SHAP) analysis, similarity maps, and atom-wise relevance scores, providing mechanistic insights and highlighting the importance of molecular symmetry and topological features in determining inhibitory potency. To support practical deployment, the final models were implemented in a freely accessible web application built with Flask and Streamlit for rapid prediction of novel compounds’ inhibitory potential. Overall, this work provides a transparent and reproducible QSAR workflow, clarifies the relative strengths of classical and deep learning approaches, and offers a practical tool to assist in the design of next-generation PDE5A inhibitors. ## Maceration, Soxhlet Extraction, and Steam Hydrodistillation of Anethum graveolens Seeds: Phytochemical Profiles and Cytotoxic Effects - Source: Molecules (journals) - Date: 2026-09-20T00:00:00Z - Categories: Docking & Screening - Authors: Christian Goldiș, Roxana Racoviceanu, Roxana Negrea-Ghiulai, Alexandra Mioc, Alexandra Prodea, Elisabeta Atyim, Tamara Maksimovic, Alexandra T. Lukinich-Gruia, Maria-Alexandra Pricop, Codruța Șoica - Journal: Molecules - DOI: 10.3390/molecules31183337 - External ID: 1adeb62bfaef771e1aca22201d465850ca857947 - Keywords: molecular docking, IC50 - Source URL: - Dashboard article: Abstract: Dill, Anethum graveolens L., is a culinary plant used in traditional medicine to treat gastrointestinal conditions; its seeds have been previously reported to exhibit antimicrobial, anti-inflammatory, hypolipidemic, anticancer and antioxidant activities. This study aimed to assess dill seed extracts and essential oil in terms of chemical composition and cytotoxic effects, combined with an exploration of the potential underlying mechanisms. Four hydroethanolic seed extracts were prepared using Soxhlet extraction or maceration and the essential oil (AGEO) was prepared by steam distillation. Extracts were physicochemically analysed via LC-MS and spectrophotometry while the AGEO composition was determined through GC-MS. Their cytotoxic effects were assessed in A375, PANC-1 and SK-OV-3 cancer cells, while HaCaT keratinocytes were used as non-cancerous control. The mechanistic investigations included immunofluorescence assay, high-resolution respirometry, network pharmacology and molecular docking. Ferulic acid and fisetin were identified as major components of the hydroethanolic extracts, while D-limonene and carvotanacetone were the main compounds in the AGEO. All tested products were revealed to induce dose-dependent cytotoxicity (with IC50 values ranging between 619.8 and 810.2 μg/mL), diminished OXPHOS efficiency and a mitochondrial uncoupling effect. Morphological changes consistent with apoptosis were recorded, particularly in PANC-1 and A375 cells. Network pharmacology suggested a multitarget mechanism for EO that involved STAT3 and HSP90AA1, while the molecular docking calculations indicated that D-limonene and carvotanacetone could be accommodated within the ATP-binding pocket of Hsp90α. ## Machine Learning-Guided Structural Optimization of High-Temperature Polybenzimidazole-Based Phosphonic Acid Proton Exchange Membranes - Source: ACS Applied Polymer Materials (journals) - Date: 2026-09-20T00:00:00Z - Authors: Cheng-Kai Hong, Chun-Yang Yu, Yong-Feng Zhou - Journal: ACS Applied Polymer Materials - DOI: 10.1021/acsapm.6c02214 - External ID: 059407ecf2eb53b48744e936635fba76ed23e690 - Keywords: denoising, random forest, molecular dynamics - Source URL: - Dashboard article: Abstract: Polybenzimidazole-based phosphonic acid proton exchange membranes (PEMs) are promising candidates for high-temperature PEM applications. However, the enormous monomer design space and the sensitivity of proton transport to local chemical environments render the rational membrane design highly challenging. In this work, a machine learning-assisted strategy integrating a denoising diffusion probabilistic model and a random forest screening model was developed to accelerate the discovery of high-performance PEM monomers. The framework generated chemically diverse candidates, identified key structural motifs governing proton conductivity, and was further validated by using molecular dynamics simulations. Mechanistic analysis revealed that enhanced proton conductivity originates from rapid hydrogen bond network reorganization and weak interactions among phosphate groups, membrane segments, and protons. This study provides an interpretable framework for the design of advanced, high-temperature proton-conducting membranes. ## Machine-learning force-field scoring rivals free-energy perturbation for congeneric ligand ranking across public benchmarks - Source: ChemRxiv (preprints) - Date: 2026-09-20T00:00:00Z - Categories: Docking & Screening, Design de novo, Free Energy & MD, ML Potentials - Authors: Kevin Ryczko, Sarah Maier, Amogh Sood, Patrick Rowe, Harish Ramadas, Lorenzo Boninsegna, Justin Overhulse, Hunter La Force, Mikayla Darrows, Shiji Zhao, Andrew Wildman, Mary Pitman, Romelia Salomon Ferrer, Andrea Bortolato - DOI: 10.26434/chemrxiv.15008810/v2 - External ID: 10.26434/chemrxiv.15008810/v2 - Keywords: Machine learned force fields, free energy perturbation, MACE, lead optimization, force fields, force field - Source URL: - Dashboard article: Abstract: Relative binding free-energy (RBFE) methods such as free-energy perturbation (FEP) set the accuracy standard for ranking congeneric ligands in structurebased drug design, but their computational cost limits throughput. Machine-learned force fields (MLFFs) now approach quantum-chemical accuracy at a small fraction of that cost, raising the question of whether they can recover much of RBFE’s ranking accuracy while remaining computationally efficient. We introduce a static-score methodology called taqo and benchmark 10 MLFF variants spanning 6 model families (UMA, MACE, Orb, eSEN, AIMNet2, AllScAIP), together with a classical molecular-mechanics force field, on a uniform staticpocket protocol, correlated against experiment across 22 benchmark systems.We find that the ten MLFFs cannot be resolved from one another at this sample size, while every one of them significantly outranks the classical force field reference, at a few seconds per ligand on a single GPU. Against published OpenFE and FEP+ results on the same systems and the same ligand sets, taqo is statistically indistinguishable from OpenFE and significantly below FEP+, at roughly three orders of magnitude lower cost. On the congeneric JACS subset taqo and FEP+ are statistically indistinguishable. On the same benchmark poses taqo outperforms classical docking scores and is statistically indistinguishable from deep-learning affinity models, without any affinity-specific training. Together these establish MLFF interaction-energy scoring as a practical high-throughput approach for lead optimization. ## MPID Polarizable Force Field: GPU Implementation and Biomolecular Simulations - Source: ChemRxiv (preprints) - Date: 2026-09-20T00:00:00Z - Categories: Free Energy & MD - Authors: Qiaozhu Tan, Andrew C Simmonett, Bernard R. Brooks, Jing Huang - DOI: 10.26434/chemrxiv.15003610/v2 - External ID: 10.26434/chemrxiv.15003610/v2 - Keywords: force field parameters, molecular dynamics, Force Field, CHARMM - Source URL: - Dashboard article: Abstract: The multipole and induced dipole (MPID) model provides a sophisticated description of electrostatics through permanent atomic multipoles up to the octopole level and point induced dipoles, but its high computational cost has limited its routine application to biomolecular simulations. Here, we present a GPU-accelerated implementation of the MPID model in OpenMM. The implementation produces energies and forces fully consistent with those obtained from CHARMM and enables routine microsecond-scale molecular dynamics simulations of explicitly solvated protein systems. Moreover, force field parameters were transferred directly from the Drude oscillator model to the MPID framework without reparameterization, yielding the MPID-2019 protein force field. Tests on five proteins show that MPID-2019 reproduces a range of NMR observables, including scalar couplings and relaxation order parameters. These results demonstrate that MPID-2019 is a valid polarizable force field readily applicable to protein simulations and that the effective equivalence of the induced dipole and Drude oscillator formalisms can be established at the biomolecular level. ## Next-Generation Biomarkers and Artificial Intelligence in Colorectal Cancer: From Multi-Omic Data Integration to Clinical Application - Source: Cancers (journals) - Date: 2026-09-20T00:00:00Z - Authors: Christina Loukopoulou, Ioannis Koliarakis, John Tsiaoussis - Journal: Cancers - DOI: 10.3390/cancers18183052 - External ID: 7efc3d3d46069f45551f4220c67f9c516a96dfd4 - Source URL: - Dashboard article: Abstract: Colorectal cancer (CRC) remains a leading cause of cancer-related mortality worldwide, although advances in molecular profiling and artificial intelligence (AI) are now reshaping precision oncology in ways that promise better patient outcomes. This review synthesises contemporary evidence on next-generation biomarkers and AI applications in CRC across multi-omic data integration, liquid biopsy, computational pathology, radiomics and clinical implementation. Several findings stand out. Integrated multi-omic approaches combining genomics, transcriptomics, proteomics, metabolomics and microbiomics outperform single-omic biomarkers for predicting prognosis and treatment response. Machine learning (ML) and deep learning (DL) models achieve clinical-grade performance for molecular biomarker prediction directly from routine histopathology, with a pooled area under the receiver operating characteristic curve (AUROC) of 0.94 for microsatellite instability (MSI) detection. Circulating tumour DNA (ctDNA) monitoring enables minimal residual disease (MRD) detection and real-time treatment guidance, postoperative ctDNA status separating a two-year recurrence-free survival of 91.1% from 50.4%; randomised evidence further shows that ctDNA-guided management can safely reduce adjuvant chemotherapy use in stage II colon cancer. Radiomics and pathomics extract prognostically significant quantitative features from imaging and histopathology, permitting non-invasive tumour characterisation, while multimodal models integrating clinical, genomic, imaging and pathological data support individualised treatment selection. MSI-high (MSI-H) status predicts exceptional immunotherapy benefit, carrying an overall survival hazard ratio of 0.35 against chemotherapy. Considerable difficulties nonetheless persist: harmonising data across institutions, generalising models to diverse populations, algorithmic bias and regulatory frameworks suited to clinical AI. Priorities for the coming years are prospective validation of AI-guided treatment algorithms, integration of spatial transcriptomics and single-cell approaches, federated learning to support multi-institutional collaboration without compromising privacy, and standardised protocols for biomarker testing and interpretation. ## Novel L-Asparaginases from the human gut microbiome: Genome mining, biochemical characterization, and in vitro anti-leukemic activity. - Source: Bioorganic chemistry (journals) - Date: 2026-09-20T00:00:00Z - Authors: Duygu Delican, Ozan Kılıçkaya, Ozkan Ozden, Yunus Ensari - Journal: Bioorganic chemistry - DOI: 10.1016/j.bioorg.2026.110550 - External ID: 49abb090960ded963d47404140be3eebc36d37fd - Keywords: virtual screening, enzyme - Source URL: - Dashboard article: Abstract: L-asparaginase is essential for acute lymphoblastic leukemia treatment; however, current Escherichia coli and Erwinia chrysanthemi formulations face significant limitations, including immunogenicity, glutaminase-associated toxicity, and short plasma half-life. The human gut microbiome represents an unexplored reservoir of therapeutic enzymes that may offer superior biocompatibility due to host-commensal co-evolution. We employed a systematic genome-mining approach to screen human gut metagenomic data for novel L-asparaginase candidates. Five candidate enzymes from the genera Bacteroides, Ruminococcus, Clostridium, and Prevotella were identified using virtual screening. These enzymes were subsequently codon-optimized and heterologously expressed in E. coli, thereby validating our computational selection strategy. Biochemical characterization revealed optimal activity at alkaline pH (8.0-9.0), robust performance at physiological temperature (37 °C), and excellent storage stability. Ruminococcus\_seq7 exhibited exceptional kinetic properties (Km = 0.53 ± 0.19 mM; Vmax = 78.6 ± 5.59 U/mg), whereas Bacteroides\_seq104 showed intermediate kinetics (Km = 2.04 ± 0.57 mM; Vmax = 75.4 ± 5.75 U/mg). These lead candidates demonstrated complementary anti-leukemic profiles: Ruminococcus\_seq7 showed broad-spectrum activity against T-cell leukemias (IC₅₀: 5.1-8.6 U/mL for Jurkat, MOLT-4, and THP-1), while Bacteroides\_seq104 exhibited remarkable potency against THP-1 cells (IC₅₀ = 0.9 U/mL) and successfully overcame resistance in REH cells (IC₅₀ = 36.9 U/mL). Both enzymes maintained >95% viability in healthy HUVEC cells. This study provides proof-of-concept for the discovery of therapeutic enzyme from the human gut microbiome. The identified L-asparaginases exhibited favorable biochemical properties, potent and selective anti-leukemic activity, and enhanced safety profiles. The absence of glutaminase activity and high biocompatibility position these gut microbiome-derived enzymes as promising biotherapeutic scaffolds for next-generation leukemia treatment, pending further optimization of substrate affinity to meet clinical standards. ## Operational insights from the front lines of OpenADMET: lessons from our blind challenges - Source: ChemRxiv (preprints) - Date: 2026-09-20T00:00:00Z - Categories: ADMET & Safety - Authors: Hugo MacDermott-Opeskin, Maria Castellanos, Jonathan A. Swain, Jenke Scheen, Sean Colby, Georgia Channing, Scott Simpkins, Robert Warnerford-Thompson, Galen J. Correy, L. Naomi Handly, Sri Kosuri, John Chodera, James S. Fraser, W. Patrick Walters - DOI: 10.26434/chemrxiv.15009105/v1 - External ID: 10.26434/chemrxiv.15009105/v1 - Keywords: ADMET - Source URL: - Dashboard article: Abstract: The gap between reported machine learning (ML) model performance in drug discovery and the experience of using these models in live discovery projects remains a major hurdle for the field, because the retrospective nature of most benchmarks makes it difficult to objectively assess progress. To address this gap, OpenADMET has hosted a series of blind challenges to benchmark and advance predictive modeling in absorption, distribution, metabolism, excretion, and toxicity (ADMET). In this Perspective, we distill the operational, strategic, and community-level lessons from our first four blind challenges, providing a practical blueprint for designing and running collaborative machine learning competitions in the life sciences. In addition, we open-source our blind challenge infrastructure and data-validation tools to lower the barrier to entry for hosting rigorous, community-driven competitions and to foster a sustainable framework for collective learning in computer-aided drug discovery. ## Potent Caspase‐1 Antagonists From Dihydropyridine Derivatives: In Vitro Fluorescence Assays and Cell Imaging - Source: ChemMedChem (journals) - Date: 2026-09-20T00:00:00+00:00 - Categories: Docking & Screening - Authors: Kittiporn Nakprasit, Thitiporn Pattarakankul, Tanapat Palaga, Thanapon Charoenwongpaiboon, Waroton Paisuwan, Kavisara Srithadindang, Rachanon Eakprachasin, Islah Muttaqin, Pattraporn Chobpradit, Sucharat Sanongkiet, Mongkol Sukwattanasinitt, Chanakan Tongsook, Anawat Ajavakom - Journal: ChemMedChem - DOI: 10.1002/cmdc.70496 - Keywords: molecular docking, MD simulations, enzyme - Source URL: - Dashboard article: Abstract: Caspase‐1 is a key marker in pyroptosis and inflammation, playing a central role in the pathogenesis of various acute and chronic inflammatory diseases. While some caspase‐1 sensing and inhibition studies have been reported, the use of DHPs for these applications remains unexplored. To address this challenge, a novel fluorescent compound, zWEHD‐NH‐DHP , was synthesized and subsequently developed as a real‐time caspase‐1 detection in vitro and in cells. Furthermore, a new class of potential caspase‐1 inhibitors was investigated through the screening of a series of nonfluorogenic DHPs, utilizing our established caspase‐1 fluorescent sensing system. Among the tested compounds, THD and FD showed caspase‐1 inhibition with IC 50 values of 2.8 ± 0.1 nM and 3.2 ± 0.1 µM, respectively. THD , which exhibited comparable potency to the well‐known inhibitor VX‐765 (3.68 nM), was selected for further investigation. Moreover, a competitive inhibition mechanism of THD at the enzyme's active site was supported by molecular docking and MD simulations. Western blot analysis of PMA‐differentiated THP‐1 cells further supported this observation by showing reduced cleaved IL‐1β levels in the presence of THD . This inhibitory effect of THD on caspase‐1 activity in a complex cell system was further confirmed by the concentration‐dependent morphological changes in BMDMs. ## Prediction of the Potential of Active Compounds in Kratom Leaves (Mitragyna speciosa) as Anti-Inflammatory Candidates through a Molecular Docking Approach to COX-1, COX-2, TNF-α, and IL-1β - Source: Journal of Pharmaceutical and Sciences (journals) - Date: 2026-09-20T00:00:00Z - Categories: Docking & Screening - Authors: N. H. Akbar, Khoirunnisa Muslimawati, Aditya Maulana Perdana Putra, Putri Helena Junjung Buih - Journal: Journal of Pharmaceutical and Sciences - DOI: 10.36490/journal-jps.com.v9i3.1847 - External ID: 92df81128ae2f6792321d41fe5c11155034dea97 - Keywords: Molecular Docking, pharmacokinetic, binding affinity - Source URL: - Dashboard article: Abstract: Inflammation is a central component of numerous pathological conditions, and the limitations of currently available anti-inflammatory therapies continue to motivate the search for structurally diverse therapeutic candidates. Mitragyna speciosa Korth. contains bioactive indole alkaloids, particularly mitragynine and 7-hydroxymitragynine, that have been associated with anti-inflammatory effects. This study evaluated the predicted binding affinity and ligand-protein interaction profiles of mitragynine and 7-hydroxymitragynine against four inflammation-related targets—cyclooxygenase-1 (COX-1), cyclooxygenase-2 (COX-2), interleukin-1 beta (IL-1β), and tumor necrosis factor-alpha (TNF-α)—using molecular docking. Protein structures were obtained from the Protein Data Bank (PDB IDs 1EQG, 3LN1, 5R87, and 3EWJ, respectively). Protein and ligand preparation and interaction visualization were performed using Discovery Studio Visualizer, docking was conducted using PLANTS, and protocol validation was assessed by redocking and root mean square deviation (RMSD) analysis in Molegro Molecular Viewer. All protocols met the prespecified RMSD acceptance criterion of ≤2.0 Å, with values of 1.57 Å for COX-1, 0.77 Å for COX-2, 1.36 Å for IL-1β, and 0.55 Å for TNF-α. Within each protein target, 7-hydroxymitragynine showed the most favorable predicted score among the tested alkaloids for COX-1 and COX-2 (−17.39 and −19.56 kcal/mol, respectively), whereas mitragynine showed the more favorable predicted score for IL-1β and TNF-α (−16.72 and −20.42 kcal/mol, respectively). These results indicate target-dependent binding behavior and provide a computational basis for further structural, pharmacokinetic, and experimental evaluation of both alkaloids as potential anti-inflammatory candidates. ## Sensitized near-infrared luminescence in sulfur-coordinated lanthanide complexes for machine learning-enhanced thermometry - Source: ChemRxiv (preprints) - Date: 2026-09-20T00:00:00Z - Authors: Mikko Rautiainen, Emily Andreato, Essi Barkas, Liyan Ming, Nikita Panov, Hannu Huhtinen, Manu Lahtinen, Petriina Paturi, Riccardo Marin, Jani O. Moilanen - DOI: 10.26434/chemrxiv.15004117/v2 - External ID: 10.26434/chemrxiv.15004117/v2 - Keywords: ab initio - Source URL: - Dashboard article: Abstract: Lanthanide (Ln 3+ ) coordination complexes featuring only Ln–S bonds have optical properties uniquely modulated by soft donor atoms. Yet, these complexes and their applications are underexplored, due to the stricter environmental control required for their typical preparation relative to complexes containing Ln–O bonds. Herein, we show that Ln-S coordination is ideally suited to generate information-rich near-infrared (NIR) photoluminescence, laying the optimal ground for machine learning-driven optical sensing. We prepared three new homoleptic complexes with general formula \[K(2.2.2-cryptand)\]\[Ln( L ) 4 \] (Ln = Nd 3+ ( 1-Nd ), Gd 3+ ( 1-Gd ), or Yb 3+ ( 1-Yb ); L = 2,6-bis-(4- tert -butylphenyl)phenyldithiocarboxylate). In all complexes, the Ln 3+ ion is dodecahedrally coordinated to eight sulfur atoms, generating a crystal field that deviates from the ideal anisotropic case and leads to substantial mixing of the crystal-field states, as revealed by magnetic measurements and ab initio calculations. Temperature-dependent photoluminescence and optical absorption measurements allowed identifying the electronic transitions and energy transfer mechanisms underpinning the complexes’ optical properties, which are dominated by ligand sensitization of the Ln 3+ emission. Survey of the photoluminescence-vs-temperature (calibration) datasets indicated the amenability of 1-Nd and 1-Yb to luminescence thermometry. Chiefly, the complex spectral variations induced by thermal changes can be comprehensively analyzed via principal component analysis, thus extracting global optical features with readout precision that is one order of magnitude higher than classical, local spectral features. More broadly, this study demonstrates that spectral complexity in molecular systems sets the necessary preconditions for advanced optical sensing supported by machine learning algorithms. ## SimSJSAlert: A Similarity-Augmented Multi-View Learning Framework with Scaffold Alerts for Drug-Induced Stevens-Johnson Syndrome Risk Assessment - Source: ChemRxiv (preprints) - Date: 2026-09-20T00:00:00Z - Categories: Property Prediction, ADMET & Safety - Authors: Huynh Anh Duy, Sastiya Kampaengsri, Supreeya Paiboon, Tarapong Srisongkram - DOI: 10.26434/chemrxiv.15008910/v2 - External ID: 10.26434/chemrxiv.15008910/v2 - Keywords: QSAR, transformer, molecular representations, molecular descriptors - Source URL: - Dashboard article: Abstract: Stevens-Johnson syndrome (SJS) is a rare, life-threatening adverse drug reaction for which early identification of high-risk drugs remains challenging. In this study, we developed an explainable quantitative structure-toxicity relationship (QSTR) framework for SJS prediction using a multi-view learning framework with 21 molecular representations, including molecular descriptors, fingerprints, transformer-based embeddings, and a molecular similarity representation. Model development and evaluation were designed in accordance with the OECD principles for QSAR validation, incorporating scaffold-aware validation, external evaluation on pharmacovigilance negative-signal compounds, and structural interpretation by integrating SHAP with Bemis-Murcko scaffold enrichment analysis. The proposed framework achieved robust sensitivity and specificity, with molecular similarity consistently improving the identification of SJS-associated compounds. Furthermore, a consensus prediction and scaffold-based prioritization strategy identified several high-risk drugs from the 1 external negative-signal dataset, including cefprozil, tetracycline, and omadacycline, highlighting their potential association with SJS and warranting further pharmacovigilance and experimental investigation. To facilitate practical application, this was deployed as a freely accessible web server, SimSJSAlert (https: //sim-sjs-alert.streamlit.app/), enabling rapid drug-induced SJS risk assessment. The proposed framework combines predictive performance, interpretability, and accessibility to support early SJS risk assessment, pharmacovigilance signal prioritization, and safer drug development. ## The Algorithm Question: When Architectural Complexity Meets Biological Reality in Antimicrobial Peptide Prediction - Source: WIREs Computational Molecular Science (journals) - Date: 2026-09-20T00:00:00+00:00 - Authors: Xingwen Long, Yikun Lin, Jiajing Xie, Jingyan Wang, Xuan Wei, Lifang Wei, Zixin Deng, Jiangtao Gao - Journal: WIREs Computational Molecular Science - DOI: 10.1002/wcms.70087 - Keywords: generative models, transformer - Source URL: - Dashboard article: Abstract: Computational prediction and design of antimicrobial peptides (AMPs) has progressed quickly, moving from physicochemical descriptors paired with classical machine learning to transformer architectures and generative models. Each new architecture claims improved performance, yet basic questions remain unanswered. Does added complexity produce proportionally better biological predictions, and what do different architectures actually learn? Recent panoramic surveys have cataloged progress; the present review takes a different stance, arguing that the field has favored architectural novelty over biological understanding and has widened the gap between computational metrics and therapeutic relevance. Three findings emerge from a re‐examination of the evidence. First, deep learning models do not consistently outperform well‐designed shallow approaches, and both encode statistically similar chemical information. Second, reported in vitro validation rates of 59% to 100% for heavily pre‐screened designs cluster within a narrow band irrespective of algorithm class, which suggests that pre‐screening pipelines, rather than raw model accuracy, drive these numbers. Third, generative models such as HydrAMP, Multi‐CGAN, and latent‐diffusion frameworks sample within the physicochemically constrained envelope defined by their training data, and because antimicrobial activity can change sharply with a single substitution, descriptor‐space novelty does not by itself establish functional novelty. To support such a critique with original evidence, the validation‐rate analysis is extended across 18 representative studies. Two community instruments are then proposed: a Biology‐First Algorithm Selection (BFAS) decision framework that operationalizes a four‐axis selection protocol over task, data, interpretability, and compute budget; and AMP‐CASP, an AMP‐equivalent of the CASP blind‐benchmark that turns an earlier call for a CASP‐like community assessment into an operational protocol, with held‐out activity, toxicity, and stability labels, homology‐aware splitting, organism‐resolved endpoints, and explicit assay‐condition metadata. Progress now depends on an honest assessment of what algorithms can and cannot deliver in moving from computational prediction to clinical translation. ## Ultrasound-Assisted Deep Eutectic Solvent Extraction of Polysaccharides: Mechanistic Foundations, Structural Consequences, and Process Optimization - Source: Polymers (journals) - Date: 2026-09-20T00:00:00Z - Authors: K. Cheong, Si Xu, Wan-Zi Yao, Farwa Abdul Hafeez, Amanullah Sabir, Afifa Aziz, Zhan-Hui Cao, Udayakumar Veerabagu - Journal: Polymers - DOI: 10.3390/polym18182298 - External ID: 57c47804220d03446d796aa36332ce3a406796b1 - Source URL: - Dashboard article: Abstract: Ultrasound-assisted deep eutectic solvent (DES) extraction has emerged as a promising green and intensified strategy for recovering natural polysaccharides from plant, algal, fungal, and other biological matrices. By coupling acoustic cavitation with tunable solvent microenvironments, this approach can enhance cell-wall disruption, solvent penetration, mass transfer, and polysaccharide solubilization while reducing reliance on harsh acidic, alkaline, or organic solvents. However, extraction efficiency alone is insufficient to define process quality because ultrasound-assisted DES systems may also reshape the molecular weight distribution, monosaccharide composition, uronic acid or sulfate content, substitution pattern, charge density, conformation, surface morphology, and physicochemical behavior. These structural consequences directly influence downstream bioactivities, including antioxidant, hypoglycemic, prebiotic, anti-inflammatory, and anti-ulcerative colitis effects. This review critically summarizes the mechanistic foundations of ultrasound–DES synergy, analyzes how extraction conditions determine polysaccharide structural outcomes, and highlights the importance of linking structure with functionality. Emerging data-driven approaches, including solvent prescreening, COSMO-RS, and machine learning-assisted process optimization, are also discussed as supporting tools for navigating the multidimensional extraction space. However, their current application to the direct prediction of polysaccharide structural outcomes remains limited. Future progress will require standardized datasets, advanced structural characterization, causal structure–activity validation, and scalable process engineering. Overall, ultrasound-assisted DES extraction should be viewed as a structure-sensitive extraction platform whose performance depends on the coordinated control of solvent properties, acoustic conditions, biomass characteristics, and downstream processing. ## Vincoside Lactam from Edible Uncaria rhynchophylla: A Food-Derived Indole Alkaloid for Dietary Postprandial Glycemic Regulation and Functional Food Development - Source: Foods (journals) - Date: 2026-09-20T00:00:00Z - Categories: Docking & Screening - Authors: Hao-Shu Liu, Hai-Rong Xiang, Huan-Nan Li, Ru-Yu Jiang, Yue Zhang, Lin-Feng Zhao, Da-Wei Zeng, Jia-Zhen Xie, Yan-Ju Gong, Xiong-Bin Chen, Lan Yang - Journal: Foods - DOI: 10.3390/foods15183329 - External ID: 428de7f36680147d326ecfeeddfe673006a88ebf - Keywords: molecular docking, enzyme - Source URL: - Dashboard article: Abstract: α-Glucosidase inhibitors play a role in postprandial blood glucose. However, synthetic hypoglycemic drugs often cause gastrointestinal discomfort and are unsuitable for use as food additives. Natural compounds from plants have therefore attracted interest as safer candidates for hypoglycemic functional foods. Vincoside lactam (VCS-LT), an indole alkaloid isolated from the leaves of Uncaria rhynchophylla (Miq.) Miq. ex Havil., is traditionally used in East Asian herbal teas and dietary supplements. We evaluated its carbohydrate-digestion inhibitory activity to provide experimental evidence for the development of herbal beverages and grain supplements with hypoglycemic properties. The in vitro assay used yeast-derived α-glucosidase, whereas the molecular docking targeted human maltase-glucoamylase (MGAM)—a distinction that is important for interpreting the respective findings. Molecular docking indicated the stable binding of VCS-LT to two catalytic domains of maltase-glucoamylase, a key intestinal enzyme involved in reducing postprandial glucose excursion after maltose loading following the consumption of high-carbohydrate staple foods. In a p-nitrophenyl-α-D-glucopyranoside-based assay using yeast-derived α-glucosidase, VCS-LT showed potent inhibitory activity. In alloxan-induced diabetic mice, oral VCS-LT reduced fasting glucose and attenuated postprandial spikes after maltose loading. In contrast to the gastrointestinal side effects commonly associated with synthetic α-glucosidase inhibitors such as acarbose, no gastrointestinal discomfort or other adverse effects were observed during the 12-day administration period. Collectively, our findings demonstrate that VCS-LT is a potent food-derived α-glucosidase inhibitor that effectively attenuates postprandial glucose excursion in diabetic mice, which supports its potential application as a functional food ingredient for daily dietary glycemic management. Nevertheless, our work remains at the proof-of-concept level, and future studies on processing stability, bioaccessibility, food-matrix interactions, sensory acceptability, and safety at food-use levels are essential before any commercial application can be envisioned. ## Virucidal activity and differential antiviral efficacy of curcuminoid preparations against porcine epidemic diarrhea virus and feline infectious peritonitis virus. - Source: Natural product research (journals) - Date: 2026-09-20T00:00:00Z - Categories: Docking & Screening - Authors: Nantawan Deeprom, Natjira Mana, P. Panichayupakaranant, M. Sukmak, Ploypailin Semkum, P. Lekcharoensuk, Sirin Theerawatanasirikul - Journal: Natural product research - DOI: 10.1080/14786419.2026.2734559 - External ID: 201703f5200bf0bbece6d5e77cf63f511141a1c0 - Keywords: Molecular docking, EC50 - Source URL: - Dashboard article: Abstract: Curcuminoids from Curcuma longa L. have exhibited antiviral activity against several human viruses; however, their comparative efficacy against animal coronaviruses remains unclear. This study evaluated the antiviral and virucidal effects of curcumin (CUR), demethoxycurcumin (DMC), bisdemethoxycurcumin (BDMC), a curcuminoid-rich extract (CRE), and its solid dispersion (CRE-SD) against porcine epidemic diarrhoea virus (PEDV) and feline infectious peritonitis virus (FIPV). Against PEDV, all compounds exhibited multi-stage antiviral activity, confirmed by cytopathic effect reduction, IPMA, and RT-qPCR. DMC was the most consistently potent (EC50 = 0.48-0.98 µg/mL), followed by BDMC and CUR (0.37-4.47, and 1.50-1.99 µg/mL, respectively), while only BDMC lacked prophylactic activity. Purified curcuminoids showed limited activity against FIPV, though this virus remained susceptible to the extract formulations. Molecular docking supported the experimental findings. Notably, CRE and CRE-SD exerted strong virucidal activity against both viruses, suggesting synergistic effects of multiple constituents. These findings highlight curcuminoid formulations as promising veterinary antivirals and disinfectants. ## Water Layering and Orientation Modulate Lanthanide Hydration in Carbon Nanotubes - Source: ChemRxiv (preprints) - Date: 2026-09-20T00:00:00Z - Categories: ML Potentials - Authors: Kailong Zhang, Jian Shi, Qing Shao - DOI: 10.26434/chemrxiv.15009104/v1 - External ID: 10.26434/chemrxiv.15009104/v1 - Keywords: MACE, molecular dynamics - Source URL: - Dashboard article: Abstract: Nanoscale confinement can reorganize water molecules and alter ionic hydration, offering a potential route to separating chemically similar ions. We use molecular dynamics simulations with the MACE-MH-1 foundation machine-learning interatomic potential to investigate the hydration of eight trivalent lanthanide ions (Ln 3+ : La 3+ , Pr 3+ , Nd 3+ , Gd 3+ , Tb 3+ , Dy 3+ , Tm 3+ , and Lu 3+ ) in carbon nanotubes (CNTs) with diameters of 1.09–1.49 nm. We first evaluate MACE-MH-1 for Ln 3+ hydration in bulk solution. The OMOL model of MACE-MH-1 reproduces the reported Ln–O distances and coordination-number trends and is therefore used for the CNT simulations. The hydration factor, which measures first-shell orientational order, varies nonmonotonically with CNT diameter and reaches a minimum in CNT(9,9), which also accommodates the most first-shell water molecules. This minimum coincides with the emergence of a central water population and reflects two combined effects: difference in preferred orientation between first-shell and surrounding water molecules and increased hydrogen-bond coupling between them. However, different Ln 3+ ions respond similarly to confinement, suggesting that CNT confinement alone may offer limited selectivity for their separation. ## When Is a Molecule a Duplicate? Identity Policy Determines What a Benchmark Audit Finds - Source: ChemRxiv (preprints) - Date: 2026-09-20T00:00:00Z - Categories: Cheminformatics - Authors: Md. Rahul Reza Roktim - DOI: 10.26434/chemrxiv.15009099/v1 - External ID: 10.26434/chemrxiv.15009099/v1 - Keywords: MoleculeNet, InChIKey, InChI - Source URL: - Dashboard article: Abstract: Reports of duplicate or redundant records in molecular machine-learning benchmarks are routinely presented as properties of the datasets. They are not. A duplicate count is the output of an operational definition of molecular identity applied to a dataset, and different defensible definitions return different answers from the same file. We audited seven MoleculeNet \[3\] benchmarks (52,144 records) under a four-level identity hierarchy — canonical molecular graph, fragment parent, charge-normalised parent and tautomer-normalised parent — with the Standard InChIKey as an orthogonal comparator rather than a rung, and with derived identity withheld outside a declared domain of applicability. Duplicate-group counts moved with the level in every benchmark, and for three the audit changed its answer to whether the benchmark contains duplicate compounds at all. Domain choice mattered as much: HIV reports 330 duplicate groups at the tautomer-normalised level when identity is asserted for every record, against 124 on the declared domain. A post hoc sensitivity analysis found that domain under-covered main-group and f-block metal-associated structures, giving 109 without displacing 124 as the prespecified result. Of 295 merge events, 179 were decidable from structure alone and 116 were policy-dependent; across that complete census, 19 events (16.4%; 41 records) received a different benchmark-equivalence decision from transition-appropriate official-InChI arbitration under an endpoint rulebook fixed before final classification, all 19 in one benchmark's stereochemical subset. We publish the identity policy as an executable, fingerprinted contract reproducing every reported count exactly (434/434), so the assumptions behind an audit statistic can be inspected and re-executed rather than inferred. ## ECENet: An Edge Cluster Expansion Line-Graph Neural Network - Source: arXiv (preprints) - Date: 2026-09-19T16:59:42Z - Categories: Property Prediction, Spectra & Analytical, ML Potentials - Authors: R. Allen LaCour, Teresa Head-Gordon - External ID: 2609.23134v1 - Keywords: MACE, equivariant, Graph Neural Network, graph neural networks, force fields - Source URL: - Dashboard article: - PDF: - Code: Abstract: Machine-learned interatomic potentials (MLIPs) have emerged as a promising alternative to classical force fields and first-principles theory for predicting the properties of chemical and material systems. Many MLIPs are graph neural networks with O(3)-equivariant features, whose accuracy comes at substantial computational cost. Here we introduce the edge cluster expansion (ECE), an analogue of the atomic cluster expansion in which the environment is expanded around edges between atom pairs rather than single atoms, and build upon it to develop the line-graph neural network ECENet. ECENet uses O(2)-equivariant features that persist on the edges between atoms, giving it natural access to O(2) operations that are cheaper and less restrictive than their O(3) counterparts. ECENet performs at the state of the art on the MD22 benchmark and lies on the accuracy-cost Pareto frontier when trained on the SPICE-MACE-OFF dataset. Furthermore, with long-range electrostatics implemented via latent Ewald summation, ECENet accurately predicts molecular dipole moments and the infrared spectrum of liquid water. The frontier ECENet architecture establishes equivariant edge-centered representations as an efficient and physically expressive foundation for equivariant MLIPs. ## MolSC: Leveraging Substituent Contributions to Enhance Fine-grained Molecular Understanding in LLMs - Source: arXiv (preprints) - Date: 2026-09-19T15:12:45Z - Categories: Property Prediction, LLMs & Agents - Authors: Hyuntae Park, Sooyeon Kim, Jiwon Park, SangKeun Lee - External ID: 2609.23073v1 - Keywords: LLMs, GPT, bioactivity - Source URL: - Dashboard article: - PDF: - Code: Abstract: Recent advances in natural language processing have led to molecular Large Language Models (LLMs) with strong performance across diverse chemistry tasks. However, they still struggle to capture fine-grained structure-property relationships, particularly how small, localized modifications alter a molecule's behavior. To address this limitation, we introduce MolSC, a dataset of substituent contributions, defined as property changes induced by attaching specific substituents to molecular scaffolds. Curated from manually annotated bioactivity records, MolSC spans structural-alert liability, target-specific bioactivity, and physicochemical descriptors, and contains 181K substituent-level examples for training. We further propose MolSC-Bench, a held-out evaluation benchmark of 1,541 examples disjoint from MolSC at the scaffold, substituent, and molecule levels. Our experiments show that existing molecular LLMs and strong proprietary models such as GPT-5.2 and Gemini-3-Flash show limited reliability in substituent contribution prediction. In contrast, training on MolSC substantially improves this ability and achieves strong performance across diverse downstream molecular tasks. These results highlight substituent contribution learning as a key component of fine-grained molecular understanding. ## CurvFlow-DTA: dual-graph discrete Ricci curvature flow for drug--target affinity prediction - Source: arXiv (preprints) - Date: 2026-09-19T08:02:34Z - Categories: Cheminformatics, Property Prediction - Authors: Jicheng Ma, Yunyan Yang, Juan Zhao, Liang Zhao - External ID: 2609.22862v1 - Keywords: SMILES, Graph neural networks - Source URL: - Dashboard article: - PDF: Abstract: Graph neural networks are widely used for drug--target affinity (DTA) prediction, and discrete Ricci curvature has recently been used to characterize molecular graph geometry. Existing curvature-aware DTA approaches mainly use static curvature on the drug graph while representing proteins primarily with sequence-derived features. This leaves pair-adaptive use of graph geometry underexplored, which may limit adaptation to unseen entities in cold-start settings relevant to practical screening. We present CurvFlow-DTA, which replaces a single static curvature representation with weighted Forman curvature flow on both molecular and protein residue--residue contact graphs. A label-independent flow trajectory is precomputed for each entity, and a pair-conditioned selector determines the horizons read by a dual-branch Flow-GINE. A frozen ESM-2 supplies residue-level representations and contact scores used to construct the protein graph. Inference requires only SMILES strings and protein sequences, without a bound complex structure. On Davis and KIBA, CurvFlow-DTA improves on the protocol-matched Ricci-GraphDTA baseline in every warm and cold-start setting. Warm-split mean squared error (MSE) decreases by $19.9\\%$ on Davis and $18.9\\%$ on KIBA. Across the six cold-start comparisons, MSE decreases by $14.3$--$27.4\\%$, with higher concordance index (CI) throughout. Within our compiled set of literature baselines, CurvFlow-DTA achieves the lowest MSE on both warm benchmarks and across four out of six cold-start evaluation settings. ## SPIBER: Reconstructing Free Energy Landscapes from Short, Unconverged Trajectories with Generative Flow Networks - Source: arXiv (preprints) - Date: 2026-09-19T00:37:45Z - Authors: Venkata Sai Sreyas Adury, Pratyush Tiwary - External ID: 2609.22663v1 - Keywords: molecular dynamics, Free Energy - Source URL: - Dashboard article: - PDF: Abstract: Molecular systems have many degrees of freedom, but their metastable behavior can often be described by a few collective variables. Identifying these variables and estimating free energies along them from limited simulation data remains a challenging, important problem. Separate short trajectories may sample different metastable states without capturing transitions or establishing their relative equilibrium populations. For unbiased trajectories generated with the same Hamiltonian at a single temperature, alternate methods based on histogram reweighting cannot correct this imbalance. Here we present SPIBER, which combines the State Predictive Information Bottleneck (SPIB) with Generative Flow Networks (GFlowNets). SPIB uses deep learning to approximate slow degrees of freedom through a past-future information bottleneck, retaining information needed to predict future metastable states. We show that this compression limits conditional entropy variations in populated regions, allowing conditional mean potential energies, which are much easier to calculate, to be used to approximate free energy differences. Given sufficient local sampling to estimate these energies, they define the target distribution for GFlowNets, energy-based generative samplers that sample according to estimated thermodynamic stability rather than observed populations. For a particle in a radial double-well potential, for alanine dipeptide, and for the nine-residue peptide AIB9, SPIBER recovers free energy differences between sampled metastable states to within one thermal energy unit of reference values. The method combines collective-variable learning and free energy estimation in up to four latent dimensions, without requiring converged state populations or additional molecular dynamics simulations. ## A microfluidic electrochemical biosensing platform for label-free cytotoxicity assessment of disinfection byproducts with mechanistic signal interpretation. - Source: Biosensors & bioelectronics (journals) - Date: 2026-09-19T00:00:00Z - Categories: Docking & Screening - Authors: Zhi-Peng Zhang, Ying Liu, Hui-Zi Zheng, Guan-Lan Wu, Xiao-Lin Zhu, Jiao Qu - Journal: Biosensors & bioelectronics - DOI: 10.1016/j.bios.2026.119237 - External ID: 471c956a3eef13f5f6c2ab92484c8c94fe744bfc - Keywords: molecular docking, binding affinity - Source URL: - Dashboard article: Abstract: Conventional in vitro cytotoxicity assays, when used as biosensing platforms, suffer from high reagent consumption, poor signal reproducibility, and low throughput, limiting their utility for generating high-quality training data for computational models. Herein, we develop a novel cell-based microfluidic electrochemical biosensing platform featuring a 9-channel PDMS/glass microfluidic chip for uniform monolayer culture, and a MWCNT-COOH/Ti3C2Tx/ionic liquid nanocomposite-modified screen-printed electrode transducer. This biosensor detects electroactive purine metabolites (including xanthine/guanine) secreted by HepG2 cells, enabling label-free monitoring of cellular metabolic status. The single-channel working volume is only 200 μL, reducing reagent and cell consumption by over 90% versus traditional systems, while the single-compound assay cycle is shortened to approximately 48 h. As a demonstration, we systematically evaluated the cytotoxicity of eight chlorinated phenylacetonitrile (CPAN) isomers. The biosensor yielded consistent dose-response results with conventional methods but with improved repeatability (RSD = 3.17%). To test whether the electrochemical signals faithfully reflect cellular stress, we performed molecular docking-assisted regression, which established a class-specific correlation between multi-target binding affinity and the biosensor's cytotoxicity readout (R2 = 0.85, Q2 = 0.79). Mechanistic analysis suggests that chlorine substitution patterns are associated with cytotoxic potency, potentially via synergistic hydrogen/halogen bond geometries, which is consistent with the differential purine metabolic disruption detected by the sensor. This work not only provides a robust and low-cost biosensing toolkit for emerging contaminants but also demonstrates a phenotype-to-mechanism workflow that bridges experimental sensing with computational insight. The modular chip design offers a flexible template for scalable screening and is readily adaptable to other contaminant classes or dynamic exposure scenarios. ## A pyridone-fused imidazo\[1,2-a\]pyridine hybrid binds to tubulin at the vinblastine site, depolymerizes microtubules and inhibits Cancer cell proliferation. - Source: Bioorganic chemistry (journals) - Date: 2026-09-19T00:00:00Z - Authors: Mehak Sood, Tuhin Sarkar, Kanchan Chaurasiya, S. C. Sahoo, Dulal Panda, P. Bharatam - Journal: Bioorganic chemistry - DOI: 10.1016/j.bioorg.2026.110552 - External ID: 613fc802b23b2664771f4974c2b750445bd69670 - Keywords: Molecular docking, DFT, ADMET, pharmacokinetic - Source URL: - Dashboard article: Abstract: Tubulin remains a key target in anticancer therapy, motivating the development of structurally novel small-molecule inhibitors. The vinblastine-binding site on tubulin is a well-established and druggable pocket that accommodates structurally diverse ligands, including small-molecule inhibitors. Imidazo\[1,2-a\]pyridine represents an important scaffold in medicinal chemistry, particularly in anticancer research. CH activation on this scaffold provides an efficient approach to access structurally diverse and bioactive fused heterocycles. Herein, we report the regioselective Ru (II)-catalyzed \[4 + 2\] C-H/alkyne annulation of imidazo \[1,2-a\]pyridine-3-methoxyamides, enabling efficient construction of pyridone-fused imidazo\[1,2-a\]pyridine frameworks which are structurally reminiscent of carbazole-based anticancer motifs. A library of thirty novel compounds of previously inaccessible species was synthesized and evaluated for anticancer activity. Several derivatives exhibited significant antiproliferative effects, with C-18 and C-28 identified as the most active against HeLa and MDA-MB-231 cell lines. Mechanistic studies demonstrated that C-18 inhibits tubulin polymerization in vitro, disrupts microtubule assembly in cells, and induces mitotic arrest. Molecular docking studies indicated that C-18 binds at the vinblastine site on tubulin, consistent with its mechanism of action. Furthermore, ADMET analysis suggested that most of the compounds possess favorable drug-like properties with acceptable pharmacokinetic profiles. SIGNIFICANCE STATEMENT: We report an efficient, oxidizing directing group-assisted CH activation strategy that overcomes severe synthetic bottlenecks to access novel pyridone-fused imidazo\[1,2-a\]pyridines. With the help of cell-free and phenotypic assays, we validate C-18 as a rare small molecule that targets the tubulin vinblastine site via competitive displacement. Validated structure and mode of formation by single-crystal X-ray data and DFT studies, C-18 drives G2/M mitotic arrest and successfully suppresses aggressive MDA-MB-231 breast cancer cell migration, offering a distinct chemical blueprint for microtubule-targeted oncology drug discovery. ## Advances in Cubosome-Based Transdermal Delivery: Structural Design, Manufacturing, and Engineering Strategies - Source: ACS Omega (journals) - Date: 2026-09-19T00:00:00+00:00 - Categories: Spectra & Analytical - Authors: Maxius Gunawan, Ahmad Efendi, Delly Ramadon, Celine Valeria Liew, Romchat Chutoprapat - Journal: ACS Omega - DOI: 10.1021/acsomega.6c07490 - Source URL: - Dashboard article: Abstract: Cubosomes are third-generation lipidic nanocarriers characterized by bicontinuous cubic liquid-crystalline nanostructures and exceptionally high internal surface areas. These architectures enable the encapsulation of hydrophilic, hydrophobic, and amphiphilic compounds while providing enhanced stability, cargo protection, and controlled drug release. Their ordered nanostructure, governed by lipid self-assembly and critical packing parameter principles, promotes an effective interaction with the stratum corneum, making cubosomes promising transdermal delivery platforms. This review summarizes the physicochemical principles, fabrication technologies, and engineering strategies underlying cubosome-based transdermal systems. Specific crystallographic mesophases are influenced by lipid composition, stabilizers, guest molecules, and fabrication conditions, which determine the structural integrity and delivery performance. Despite their advantages, conventional cubosomes still exhibit limitations, including high bulk viscosity, burst release, inadequate deep dermal penetration, phase instability, and limited long-term stability. To overcome these challenges, advanced approaches including chemical phase modulation, hybrid integration with microneedles or hydrogels, surface functionalization, and stimuli-responsive systems are discussed. Emerging technologies such as microfluidics, quality by design, process analytical technology, artificial intelligence/machine learning-assisted optimization, and smart personalized transdermal drug delivery systems are also highlighted for next-generation cubosome development. These integrated strategies improve skin permeation, formulation stability, therapeutic efficacy, and manufacturing scalability while supporting the future translational development of cubosome-based nanomedicines for transdermal administration of small molecules and biomacromolecular therapeutics. ## An Ensemble Graph-Transformer/Descriptor-Fusion Model with Uncertainty Diagnostics for Ternary Class Screening of Daphnia magna Acute Toxicity. - Source: Environmental pollution (journals) - Date: 2026-09-19T00:00:00Z - Categories: Property Prediction, ADMET & Safety - Authors: Ze-Xin Wen, Rong Yang, Rui-Dong Chen, Yuxuan Chen, Yu-Yao Jiang, Qing-Yi Cao, Shu-Ying Li, Hai Yu, Wen-Jun Gui - Journal: Environmental pollution - DOI: 10.1016/j.envpol.2026.129200 - External ID: f3fcdbc7b7b15d3dd05ea42f41b95652283e5a0f - Keywords: QSAR, classification model, Transformer, chemicals - Source URL: - Dashboard article: Abstract: Accurately predicting the acute toxicity of chemicals to Daphnia magna is critical for environmental risk assessment. Yet, current deep learning models often exhibit limited generalization and overconfidence when processing out-of-distribution molecules, risking dangerous false-negative errors due to scarce training data. To tackle this issue, the present study proposes GTFCN (Graph Transformer and Fully Connected Network), a ternary classification model within a two-branch architecture. One branch combines graph convolution with a masked Transformer to capture local atomic environments and higher-order interactions, while the other encodes ten global physicochemical descriptors through a fully connected network. The two representations are adaptively fused via a learnable gate, and a deep ensemble uncertainty framework is incorporated to quantify predictive uncertainty. Using a newly constructed dataset of 1,801 chemical compounds, GTFCN achieved 70.6% accuracy on a test set (181 compounds) and 71.4% accuracy on an external set (28 compounds). Crucially, the uncertainty framework effectively mitigated model overconfidence. With a threshold of H\* = 0.633 selected exclusively on the validation set and subsequently fixed for target evaluation. On the external set, the low-uncertainty subset retained by this threshold comprised 19 compounds, of which 16 were correctly classified (84.2% accuracy). This research not only overcomes the "black-box" limitations of traditional QSAR models but also provides a practical tool for early-stage screening and prioritization of chemicals with potential acute toxicity to D. magna. ## Artificial Intelligence for Predictive Mixture Toxicology. - Source: Toxicology (journals) - Date: 2026-09-19T00:00:00Z - Authors: J. Domingo, Marília Cristina Oliveira Souza, Fernando Barbosa - Journal: Toxicology - DOI: 10.1016/j.tox.2026.154589 - External ID: 5d57c0ef8550b7c0abe81d83e2eea186fc8f5b30 - Keywords: graph neural networks, molecular descriptors, chemicals - Source URL: - Dashboard article: Abstract: Human populations and ecosystems are continuously exposed to complex mixtures of environmental contaminants rather than to individual chemicals in isolation. These mixtures include pesticides, metals and metalloids, persistent organic pollutants, endocrine-disrupting chemicals, per- and polyfluoroalkyl substances, pharmaceuticals, plastic-associated compounds, air pollutants, nanomaterials, and numerous poorly characterized substances. Their combined effects may be additive, synergistic, or antagonistic, and are strongly influenced by dose, component ratio, timing, exposure sequence, and biological susceptibility. Experimental evaluation of all environmentally relevant mixtures is infeasible because the number of possible combinations increases combinatorially. Artificial intelligence (AI) offers a possible way to address this limitation. Machine learning, deep learning, graph neural networks, Bayesian approaches, and natural-language processing can integrate heterogeneous data, including chemical structures, molecular descriptors, toxicokinetics, high-throughput screening results, omics profiles, adverse outcome pathways, biomonitoring data, and epidemiological findings. However, the evidence base is uneven. Relatively few studies have applied AI directly to experimentally characterized mixtures, and much of the current optimism is extrapolated from single-chemical toxicology. This review distinguishes explicitly between applications demonstrated in mixtures, proof-of-concept mixture applications, and approaches whose mixture use remains prospective. It further examines the methodological requirements for mixture prediction, including dose and ratio representation, additivity reference models, applicability domains, external validation, and mechanistic interpretability, and proposes a framework for regulatory-grade implementation. Current evidence does not support autonomous AI-driven regulation of mixtures. AI should complement, not replace, experimental and expert evaluation, supporting a transition toward more predictive, mechanism-informed assessment of real-world chemical exposures. ## Bridging Nature, Data, and Artificial Intelligence: An Interdisciplinary Approach for Natural Products Discovery - Source: Science and Technology Nexus (journals) - Date: 2026-09-19T00:00:00Z - Categories: Property Prediction, LLMs & Agents - Authors: N. Hegazi - Journal: Science and Technology Nexus - DOI: 10.25259/stn\_11\_2026 - External ID: a5b60cc16ead04b7199db83f40836f9ec7cdb3a2 - Keywords: virtual screening, graph neural networks, GNNs, synthesis planning, bioactivity - Source URL: - Dashboard article: Abstract: Natural products (NPs) continue to serve as a foundational source for drug leads, owing to their unique chemical architectures, evolutionary optimisation, and structural diversity. Nevertheless, conventional natural product discovery is slow, resource-intensive, and constrained by rediscovery, dereplication inefficiencies, silent biosynthetic gene clusters (BGCs), complex structural elucidation, and insufficient integration of genomic, spectral, chemical, and biological data. The integration of multiple disciplines in NPs discovery creates extraordinary opportunities as well as significant challenges in heterogeneous data integration and interpretation. Recent advances in artificial intelligence (AI) and its tools such as machine learning (ML), deep learning (DL), graph neural networks (GNNs), chemical language models, foundation models, multimodal learning frameworks, knowledge graphs, and emerging agentic AI systems are reshaping NPs research, enabling the analysis and integration of huge and heterogeneous datasets. Their applications extend from genome and BGCs mining to metabolite annotation, spectral interpretation, dereplication, bioactivity and target prediction, virtual screening, de novo natural product-inspired design, and synthesis planning. Despite its potential, the full integration of AI in the NPs discovery pipeline is still limited owing to data scarcity, inconsistent annotations, fragmented databases, limited interoperability among community resources, and the lack of standardised metadata frameworks. This review highlights different AI technologies and their applications through the NPs pipeline with a focus on multi-omics integration and the growing ability of AI to connect biosynthetic potential with chemical structures, biological functions, and therapeutic relevance. Further current challenges and future directions toward a closed-loop AI-driven discovery pipeline that integrates computational prediction, experimental validation, and active learning are discussed. ## Computational designing of multiepitope based vaccine (MEBV) candidate against human metapneumovirus (HMPV) using immunoinformatics approach - Source: Scientific Reports (journals) - Date: 2026-09-19T00:00:00+00:00 - Categories: Docking & Screening - Authors: Sajid Naeem, Amna Rasheed, Aqsa Khan, Anas Azhar, Sajjad Ullah, Seerat Fatima - Journal: Scientific Reports - DOI: 10.1038/s41598-026-67755-9 - Keywords: Molecular docking, MD simulations - Source URL: - Dashboard article: Abstract: Human metapneumovirus (HMPV) is a paramyxoviridae family respiratory pathogen, which is closely related to respiratory syncytial virus. It produces the symptoms of the flu and it may result in the serious damage of the airways and even though it is a worldwide problem, there is no vaccine available as yet. Three antigenic proteins namely surface glycoprotein, nucleoprotein and phosphoprotein were screened from all the receptors and showed no similarity to host genome. The three viral proteins were then screened for identifying the cytotoxic and helper T cell lymphocytes as well as linear and conformational B-cell lymphocytes. The vaccine had 13 MHC-I epitopes, one MHC-II epitope, and four linear B-cell epitopes, which were combined with linkers and beta-defensin as adjuvant into a 296-amino-acid MEBV with antigenicity score of 0.6644. Molecular docking and MD simulations showed that there is a high degree of interaction with human TLR3 with binding energy of − 14.6 and immune activation. Moreover, the disulfide engineering enhanced construct stability. Cloning and codon optimization was used to verify efficient expression in the target host. Taken together, these findings suggest that the proposed MEBV has potential as an immunogen of HMPV. ## Discovery of natural product antagonists for the P2X7 receptor through virtual screening. - Source: Biomedicine & pharmacotherapy = Biomedecine & pharmacotherapie (journals) - Date: 2026-09-19T00:00:00Z - Authors: N. C. da Silva Ferreira, L. Lima, Nicole de Menezes Macedo, Esther dos Santos Campos, A. P. Alberto, Lucas G. Viviani, A. D. Do Amaral, L. A. Alves - Journal: Biomedicine & pharmacotherapy = Biomedecine & pharmacotherapie - DOI: 10.1016/j.biopha.2026.119928 - External ID: 6e4f7abd22dc986f5dede792a5bfb02d673a48ca - Keywords: drug likeness, Lipinski, virtual screening, binding affinity, receptor - Source URL: - Dashboard article: Abstract: The P2X7 receptor is a purinergic ionotropic receptor widely expressed in hematopoietic and glial cells. P2X7 receptor activation triggers a cascade of proinflammatory responses and cell death, thereby contributing to the pathogenesis of inflammatory disorders, neurodegenerative diseases, and pain. Although in vivo studies of P2X7 receptor antagonists have shown significant anti-inflammatory and antinociceptive effects, no antagonist has yet been approved for clinical use, highlighting the need for novel inhibitors. In this study, virtual screening was employed to identify natural product-derived candidates with potential P2X7 receptor antagonistic activity. Approximately 21,000 compounds from the ZINC15 database were first evaluated by docking, and the top 100 were ranked based on binding affinity scores ranging from -14.109 to -11.346 kcal/mol. Notably, these values surpassed those of most reference P2X7 receptor ligands used as controls. Five compounds (C28, C58, C63, C68, and C71) were selected for further evaluation based on drug-likeness criteria, including agreement with Lipinski's Rule of Five. A visual inspection analysis of the best-scored docking poses revealed that all five compounds interact with key residues at the P2X7 receptor allosteric site, particularly L95, P96, L97, and Y295, through hydrophobic and hydrogen-bonding interactions. In vitro assays revealed that some compounds were noncytotoxic; however, only compound C71 significantly inhibited P2X7 receptor-mediated dye uptake and ATP-induced intracellular reactive oxygen species (ROS) formation. These findings suggest that compound C71 acts as a promising P2X7 receptor inhibitor capable of blocking ATP-induced pore formation and downstream oxidative signaling, supporting further investigation as a potential therapeutic agent. ## Discovery of Novel Peptides as DPP-IV Inhibitors: In Silico Screening, Molecular Dynamics Simulations, and Preliminary In Vitro Evaluation - Source: ACS Omega (journals) - Date: 2026-09-19T00:00:00+00:00 - Authors: Burak Calis, Fernando Berton Zanchi, Beyza Canakcimaksutoglu, Ahmet E. Yetiman, Ozkan Fidan - Journal: ACS Omega - DOI: 10.1021/acsomega.6c08022 - Keywords: molecular docking, Molecular Dynamics, pharmacokinetic, ADMET, receptor, binding affinity, free energy - Source URL: - Dashboard article: Abstract: Type 2 diabetes mellitus (T2DM) is the most common metabolic disorder characterized by impaired insulin secretion, which leads to obesity. The incretin hormone glucagon-like peptide-1 (GLP-1) has a significant role in glucose homeostasis, but its native form in the human body is highly susceptible to proteolytic degradation by dipeptidyl peptidase IV (DPP-IV). In this study, a peptide library was constructed from antimicrobial peptide databases to discover novel DPP-IV inhibitors based on structural similarity to that of native GLP-1. A total of 26 candidate peptides were shortlisted and screened using in silico analysis, including molecular docking, pharmacokinetic profiling, and molecular dynamics simulations. Particularly, two of the candidate antimicrobial peptides showed potent binding affinity to DPP-IV, with free energy scores of −15.6 and −16.5 kcal/mol, surpassing the binding affinity of native GLP-1 for DPP-IV. These peptides also exhibited lower predicted binding affinities for the GLP-1 receptor (GLP-1R) than for DPP-IV, suggesting a potential selective inhibitory effect and, consequently, a lower predicted likelihood of interfering with native GLP-1 signaling. Pharmacokinetic characteristics of the peptides were promising with an optimal absorption, distribution, metabolism, excretion, and toxicity (ADMET) profile, along with prolonged half-life and stability. Because of its better membrane permeability and more favorable predicted DPP-IV/GLP-1R binding preference, peptide 3 was further evaluated for downstream experimental validation. The biological activity test showed that peptide 3 successfully exhibited a 36.2% inhibition rate against DPP-IV in vitro. These findings indicate that both peptides 3 and 10 are promising DPP-IV inhibitor candidates. ## Distribution and Catalytic Mechanism of a Novel Carboxylesterase Degrading Multiple Aryloxyphenoxypropionate Herbicides in Bacteria and Archaea - Source: Journal of Agricultural and Food Chemistry (journals) - Date: 2026-09-19T00:00:00Z - Categories: Docking & Screening - Authors: Yan-Mei Liu, Xiao-Tian Cao, Yi-Yun Chen, Ying-Ying Bao, Xu-Ke Pan, Xiao-Yun Liu, Meng-Hao Li, Feng Zhao, Wen-Feng Gong, Ning Lv, Hui-Hua Tan - Journal: Journal of Agricultural and Food Chemistry - DOI: 10.1021/acs.jafc.6c07577 - External ID: 32734136e27fc931df9c701d542aeb5743090020 - Keywords: Molecular docking, binding affinity - Source URL: - Dashboard article: Abstract: Aryloxyphenoxypropionate (AOPP), a widely used class of herbicides, pose significant risks to ecosystems and human health. In this study, Bacillus amyloliquefaciens L1 was isolated as a novel degrader, completely degrading 97.42 mg/L cyhalofop-butyl within 12 h via ester bond hydrolysis to cyhalofop acid. A new carboxylesterase, PnbA, was identified from strain L1, showing substrate preference for cyhalofop-butyl > clodinafop-propargyl > haloxyfop-p-methyl > quizalofop-p-ethyl. Molecular docking and mutagenesis identified S190 as a key catalytic residue. Two mutants, K411A and T412A, exhibited enhanced relative activities toward both cyhalofop-butyl (124.08 and 153.16%, respectively) and quizalofop-p-ethyl (120.97 and 171.06%, respectively) due to accelerated substrate capture and improved binding affinity. Strain L1 effectively remediated cyhalofop-butyl from contaminated soil and water matrices in natural microbial backgrounds. Homologous protein analysis revealed an overwhelming bacterial origin (99.39%) versus archaea (0.61%). This study provides novel enzymatic resources for AOPP remediation and expands the understanding of its microbial genetic basis. ## Enzymatic preparation of a sciadonic acid-enriched phospholipid fraction and its in-vitro lipid-accumulation-inhibitory effect. - Source: Food chemistry (journals) - Date: 2026-09-19T00:00:00Z - Categories: Docking & Screening - Authors: Qin Ye, Si-Yuan Chen, Shu-Yi Wang, Xianghe Meng - Journal: Food chemistry - DOI: 10.1016/j.foodchem.2026.151194 - External ID: b0f72d7a46d624f0521a44931b66aa40085c79a3 - Keywords: Molecular docking, enzyme - Source URL: - Dashboard article: Abstract: Sciadonic acid (SCA) is an unusual Δ5 polyunsaturated fatty acid reported to affect lipid metabolism and inflammation. Because phospholipid forms of bioactive fatty acids may improve bioavailability and modify biological activity, this study prepared SCA-enriched phospholipid fractions. SCA ethyl ester (SCA-EE) derived from Torreya grandis seed oil was transesterified with phosphatidylcholine (PC) in \[C₄MIM\]Tf₂N using Novozym 435. Under optimized conditions (PC/SCA-EE molar ratio 1:7, 20% enzyme loading, 55 °C, 24 h), SCA incorporation into the purified PC fraction reached 25.96%, as determined by GC/FID. Molecular docking and FTIR were used to investigate possible lipase-substrate interactions and lipase secondary-structure changes in ionic-liquid systems, respectively. In an oleic-acid-induced HepG2 model, the SCA-enriched phospholipid fraction inhibited intracellular lipid accumulation more strongly than PC and SCA-EE at the tested mass doses. These results support ionic-liquid-assisted enzymatic transesterification as a feasible laboratory-scale approach for preparing SCA-enriched phospholipid fractions with improved lipid-accumulation-inhibitory activity. ## Exploring the role of oxidative defects in polyethylene binding and degradation by a Rhodococcus opacus R7 multicopper oxidase: a computational perspective. - Source: Journal of biological inorganic chemistry : JBIC : a publication of the Society of Biological Inorganic Chemistry (journals) - Date: 2026-09-19T00:00:00Z - Categories: Docking & Screening - Authors: L. Callea, Carla Orlando, Claudio Greco, Gioele Fumagalli, Andrea Fasano, L. De Gioia, F. Arrigoni, L. Bertini - Journal: Journal of biological inorganic chemistry : JBIC : a publication of the Society of Biological Inorganic Chemistry - DOI: 10.1007/s00775-026-02173-w - External ID: 8eed10736f791b570c8de23998ebfe5c02712704 - Keywords: molecular docking, DFT - Source URL: - Dashboard article: Abstract: The enzymatic degradation of polyethylene (PE) by oxidative enzymes remains poorly understood at the molecular level, limiting the development of effective biotechnological strategies for plastic valorization. Here, we investigate the molecular basis of PE oxidation by the laccase-like multicopper oxidase LMCO2 from Rhodococcus opacus R7 with combined DFT and molecular docking calculations. LMCO2 is a particularly relevant model system because it has been experimentally shown to oxidatively modify low-density polyethylene without the need for mediators, despite its relatively low redox potential. DFT calculations indicate that direct oxidation of aliphatic C-H bonds in pristine PE is energetically prohibitive, and that hydroxylated defects do not significantly facilitate substrate activation. In contrast, oxidation at carbon atoms adjacent to carbonyl groups is associated with substantially lower barriers, owing to stabilization of the resulting radicals through conjugation and keto-enol tautomerism. Alternative pathways involving alkoxyl radicals and β-scission were also examined and found to be energetically unfavorable. Molecular docking calculations complement this picture by showing that both pristine and pre-oxidized PE segments can access the main cavity adjacent to the T1 copper site, although binding is weak and largely non-specific. Taken together, these results support a mechanism in which LMCO2 selectively acts on pre-oxidized, carbonyl-containing regions of polyethylene, promoting further radical chemistry that may ultimately lead to chain scission and the formation of low-molecular-weight oxygenated products. ## Floral dynamics in the protandrous flowers of Bauhinia variegata (Fabaceae): phase-specific morphology, scent, and nectar production - Source: Planta (journals) - Date: 2026-09-19T00:00:00Z - Categories: Docking & Screening - Authors: Andrews Vinicius Santos da Silva, V. de Freitas Mansano, Dalton Guimarães Veloso, Ana Claudia F. Amaral, Jefferson D. da Cruz, Fabricio de Oliveira Silva, W. S. M. Fernandes, J. R. de Andrade Silva, P. Bergamo, Leandro Freitas, J. V. Paulino - Journal: Planta - DOI: 10.1007/s00425-026-05146-0 - External ID: 595334a0114eeda1c7ded6f8e9aa6f8cfc231d4c - Keywords: molecular docking - Source URL: - Dashboard article: Abstract: Protandry in Bauhinia variegata is expressed through coordinated changes in floral morphology, scent composition, secretory activity, and nectar presentation, whereas petal reflectance remains stable; molecular docking further indicates that key phase-associated volatiles are structurally compatible with a bee odourant-binding protein, supporting mechanistic plausibility of olfactory detectability without demonstrating behavioral attraction. Bauhinia variegata exhibits protandrous flowers in which floral morphology, scent production, and nectar presentation change through anthesis. Here, we examined anthesis, stigma receptivity, petal reflectance, volatile organic compound (VOC) composition across floral organs and phases, secretory structures, nectar dynamics, and the structural compatibility of selected phase-discriminant VOCs with a honeybee odourant-binding protein. Flowers exhibited prolonged anthesis, opening at night and remaining functional for 2–3 days. Pollen release predominated during the 1st day (staminate phase), whereas maximal stigma receptivity occurred on the 2nd day (pistillate phase), accompanied by elongation and curvature of the style. Petal reflectance remained stable across phases. Floral scent was spatially heterogeneous among organs and temporally dynamic across anthesis, with terpene-rich bouquets dominated by α-pinene and β-pinene and with phase- or organ-informative compounds including nonanal, limonene, trans-caryophyllene, and 6-methyl-5-hepten-2-one. Cavitated secretory trichomes and sepals showed evidence of terpene-associated secretion. Nectar volume and total sugar content peaked during the pistillate phase, coinciding with strongest stigma receptivity. Molecular docking with AmelOBP14 showed that key phase-associated VOCs are structurally compatible with a bee odourant-binding protein, providing complementary mechanistic support for their potential olfactory detectability. Together, these results show that protandry in B. variegata is expressed through coordinated changes in morphology, scent, secretory activity, and nectar presentation, indicating a complex signalling strategy; however behavioural responses of floral visitors remain to be tested. ## Genetically supported targets and drug repurposing for sleep disorders: a Mendelian randomization systematic study. - Source: Naunyn-Schmiedeberg's archives of pharmacology (journals) - Date: 2026-09-19T00:00:00Z - Categories: Docking & Screening - Authors: Zi-Yu Wang, Miao Zhang, Cong Liu, Yi-Ying Zhang, Rong Cheng, Suli Zhang, Bin Yang, Li Wang, Jun-Hong Guo, Xiao-Hui Wang - Journal: Naunyn-Schmiedeberg's archives of pharmacology - DOI: 10.1007/s00210-026-05926-1 - External ID: 3ff4d3bea61f62beb6fa6b0108fc932a9506f2f4 - Keywords: molecular docking, molecular dynamics - Source URL: - Dashboard article: Abstract: Sleep disorders present major public health issues because of their associations with chronic diseases. Despite extensive research efforts, the underlying biological mechanisms remain inadequately understood. We performed bidirectional two-sample Mendelian randomization (MR) to investigate relationships between sleep disorders and common brain diseases, followed by summary-data-based Mendelian randomization (SMR), heterogeneity in dependent instruments (HEIDI), and colocalization analyses integrating protein quantitative trait locus (pQTL) and expression quantitative trait locus (eQTL) data to prioritize genetically supported candidate genes. Compound-gene associations were subsequently evaluated, and selected clinically used compounds underwent exploratory molecular docking and molecular dynamics simulation. Two-step MR mediation analyses and MR phenome-wide association study (MR-PheWAS) were used to investigate potential mediating traits and broader phenotypic associations. Multi-omic genetic analyses prioritized six candidate genes-BCAN, HDGF, MEPE, PAM, PRR4, and IGLON5-with BCAN, HDGF, and MEPE showing the strongest genetic support. Compound prioritization retained four clinically used drugs associated with the candidate genes for structure-based assessment. Exploratory molecular docking yielded the most favorable predicted Vina score for the BCAN-danazol pair, and molecular dynamics simulation showed persistent local BCAN-danazol contacts despite substantial global conformational rearrangement of BCAN during the trajectory. Mediation analyses identified several risk factors, cerebrospinal fluid metabolites, and structural connectivity measures showing statistical evidence consistent with partial mediation of selected gene/protein-sleep disorder associations. MR-PheWAS further characterized broader phenotypic associations of the genetically supported genes. Integrating genetic and multi-omic evidence identified candidate molecular targets and compound-gene associations relevant to sleep disorders. These findings provide hypotheses for subsequent functional and pharmacological validation. ## Integration of network pharmacology, structure-based pharmacophore modeling, DFT, molecular docking, and molecular dynamics simulations to identify multi-target small molecule inhibitors against Alzheimer's disease. - Source: Naunyn-Schmiedeberg's archives of pharmacology (journals) - Date: 2026-09-19T00:00:00Z - Categories: Docking & Screening - Authors: D. O. Onwu, R. Oria, Y. A. Adekunle, M. D. Adams, O. N. Ani, Habiganuchi Adele, Terhide Samuel Tyohemba, E. Osioma, Biodun Mayowa Popoola, Oluwafemi Shittu Bakare, V. Okon, Chika Collins Maxwell - Journal: Naunyn-Schmiedeberg's archives of pharmacology - DOI: 10.1007/s00210-026-05824-6 - External ID: f0ed33806165dd15b88f679967035df1a6772ccf - Keywords: pharmacophore modeling, molecular docking, molecular dynamics, DFT - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Mechanistic Interpretability of Fine-Tuned Protein Language Models for Nanobody Thermostability Prediction - Source: Bioinformatics (journals) - Date: 2026-09-19T00:00:00+00:00 - Authors: Taihei Murakami, Yuki Hashidate, Yasuhiro Matsunaga - Journal: Bioinformatics - DOI: 10.1093/bioinformatics/btag685 - Keywords: Free Energy Perturbation - Source URL: - Dashboard article: - Code: Abstract: Motivation While Protein Language Models (PLMs) fine-tuned on biophysical data achieve high predictive accuracy, the physical principles underlying their predictions remain obscure. Deciphering these representations offers a unique opportunity to not only interpret model decisions but also to discover novel biophysical insights governing protein properties. Here, we present a framework using Sparse Autoencoders (SAEs) to extract mechanistic knowledge from PLMs fine-tuned for nanobody thermostability. Results We fine-tuned the ESM-2 model on the nanobody thermostability dataset, achieving superior performance compared to significantly larger state-of-the-art models. SAE analysis successfully decomposed the model's dense embeddings into sparse, interpretable features without loss of predictive accuracy. We characterized these features through both global and local analyses. Global analysis provided an aggregate map of position-dependent feature contributions, whereas local analysis identified specific residue-level patterns, including known determinants such as the VHH-tetrad and critical disulfide bonds, as well as candidate stabilizing residues. Free Energy Perturbation calculations supported the structural plausibility of selected residue-level hypotheses. These results show that SAE-based interpretation can generate testable, structurally grounded hypotheses for rational protein engineering. Availability The data and source code of the proposed method are available at GitHub (https://github.com/matsunagalab/paper\_nanobody-thermostability-sae) and Zenodo (DOI: 10.5281/zenodo.18012027). Supplementary information Supplementary data are available at Bioinformatics online. ## Non-muscle proteins alleviate 118 °C sterilization-induced gel quality deterioration by non-covalently modulating myofibrillar protein conformation and aggregation. - Source: Food chemistry (journals) - Date: 2026-09-19T00:00:00Z - Categories: Docking & Screening - Authors: Wenming Wu, Fan Chen, Pei Gao, Dawei Yu, Q. Jiang, E. Liao, Wen-Shui Xia - Journal: Food chemistry - DOI: 10.1016/j.foodchem.2026.151222 - External ID: 6782f8059602e1ca571484acdcc82fadf84e541e - Keywords: Molecular docking - Source URL: - Dashboard article: Abstract: The improvement effects of non-muscle proteins (soy protein isolate (SPI), egg white protein (EWP) and whey protein concentrate (WPC)) on the quality deterioration of high-temperature (118 °C) sterilized surimi gels were systematically compared. Supplementation with 4% SPI, 4% EWP, or 8% WPC effectively alleviated the heat-induced quality deterioration of gels, among which 4% SPI exhibiting the best improvement effect, increasing gel strength by 49.15% and water holding capacity by 7.6% compared with the control group. Multi-spectral analysis showed SPI improved myofibrillar protein (MP) solubility, promoted protein unfolding, and strengthened molecular cross-linking. Rheology indicated increased storage modulus (G'), while microstructural images confirmed denser gel network. SPI also inhibited excessive aggregation of MP at 118 °C. Molecular docking and dynamics simulations confirmed stable MP-SPI complexes through non-covalent interactions. These findings demonstrate innovative application of SPI in sterilized surimi gels, providing new insights for the development of ready-to-eat surimi products. ## REAPS: An All-Atom Receptor-Aware Geometric Deep Learning Framework for De Novo Design of Linear and Macrocyclic Peptide Binders - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-19T00:00:00+00:00 - Categories: Property Prediction - Authors: Yongkang Qiu, Jingxin Qiao, Lengjing Zhu, Le Du, Shengyong Yang, Jun Zou - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c02057 - Keywords: graph neural network, Receptor - Source URL: - Dashboard article: - Code: Abstract: Peptide therapeutics occupy a critical niche between small molecules and biologics, yet the sequence design of peptide binders remains challenging because their bioactive conformations and interfacial fitness are dictated by the receptor microenvironment. Current computational approaches largely treat this as a generic inverse folding problem that relies on backbone geometry alone, overlooking the detailed physicochemical constraints of the receptor pocket. To bridge this gap, we introduce the REceptor-Aware Peptide Sequence Designer (REAPS), a geometric graph neural network that reframes sequence design for peptide binders by treating the receptor as a fully observable all-atom context. Multidimensional evaluations show that REAPS outperforms the widely used inverse folding model ProteinMPNN in sequence recovery, structural fidelity, and interface-level biophysical metrics across both linear and macrocyclic peptide binders. We further integrate REAPS with structural hallucination in a closed-loop de novo discovery pipeline, yielding peptide candidates with experimentally verified functional activity at the neurokinin-3 receptor (NK3R). The source code, model checkpoints, processed data sets, preprocessing scripts, and an example design workflow are publicly available at https://github.com/Mistletoe-git/REAPS. ## Substituted methyl acridone-carboxylate derivatives as CD1/CD2 dual-targeting HDAC6 inhibitors to ameliorate the pathological phenotype of Alzheimer's disease. - Source: European journal of medicinal chemistry (journals) - Date: 2026-09-19T00:00:00Z - Categories: Docking & Screening - Authors: Jia-Nan Ye, Si-Jia Yin, Yu-Tao Lin, Shu-Hui Yang, Lie-En Ma, Yang Wang, Kang-Yang Gao, Yu-Le Wang, Xin-Hang Yang, Zhi-Chao Yang, Ning Wang, Hao Liu, Shu-Jun Xu, Wei Cui, Bin Zhang - Journal: European journal of medicinal chemistry - DOI: 10.1016/j.ejmech.2026.119350 - External ID: 44d74e53772abf38cb3b24d5c117276605c3836c - Keywords: Molecular docking - Source URL: - Dashboard article: Abstract: Histone deacetylase 6 (HDAC6) has emerged as a promising target for Alzheimer' disease (AD). Although most reported HDAC6 inhibitors were designed to target CD2 catalytic domain, recent studies highlighted the functional importance of CD1 domain for the regulation of AD-related substrates. In this study, a series of substituted methyl acridone-carboxylate derivatives bearing a methyl ester moiety were discovered as CD1/CD2 dual-targeting HDAC6 inhibitors. The leading compound 6c could direct bind to HDAC6, and selectively inhibits HDAC6 with a high affinity in vitro. Molecular docking analysis revealed that 6c forms non-chelating coordination interactions with the zinc ions in both CD1 and CD2 domains of HDAC6. 6c inhibited Aβ oligomer-induced microtubule depolymerization and microglial phagocytic impairments in vitro. Furthermore, 6c effectively prevented HDAC6-driven α-tubulin and Hsp90 deacylation, as well as cognitive impairments in Aβ1-42 oligomer-treated mice. Combined with the acceptable physicochemical properties and promising biosafety of 6c, this study suggested that substituted methyl acridone-carboxylate derivatives, as CD1/CD2 dual-targeting HDAC6 inhibitors, might be developed as a novel lead drug for the treatment of AD. ## Thermodynamic–molecular interdependencies and causal determinants governing pharmaceutical solubility in SC-CO2 systems - Source: Scientific Reports (journals) - Date: 2026-09-19T00:00:00+00:00 - Authors: Wael A. Mahdi, Adel Alhowyan, Ahmad J. Obaidullah - Journal: Scientific Reports - DOI: 10.1038/s41598-026-71535-w - Keywords: gradient boosting - Source URL: - Dashboard article: Abstract: Accurate prediction of pharmaceutical solubility in supercritical CO₂ (SC-CO₂) systems is critical for green drug formulation and process intensification, yet existing machine learning studies largely prioritize predictive accuracy while overlooking mechanistic interpretability and causal understanding. This study proposes an integrated interpretable artificial intelligence framework for modeling drug solubility in SC-CO₂ by combining high-performance gradient boosting algorithms, explainable machine learning, Bayesian optimization, and causal inference. A curated dataset comprising 1,618 pharmaceutical compounds characterized by molecular weight, melting point, temperature, and pressure was employed. Model interpretability was investigated through SHAP analysis and partial dependence plots to quantify feature contributions and nonlinear response patterns. To move beyond correlation-based interpretation, a domain-constrained causal graph together with input-perturbation (what-if) sensitivity analysis was used to compare predictive importance against hypothesized direct causal contribution, within the limits of an observational, non-experimentally-validated causal structure. Explainability analysis identified pressure, and molecular weight, -specific properties as dominant predictive drivers, while causal analysis revealed melting point as a hidden but structurally influential determinant despite its lower statistical importance. Counterfactual simulations further exposed nonlinear and context-dependent solubility responses under ± 20% perturbations of key variables, highlighting asymmetric system sensitivities. Additionally, interaction analysis uncovered strong thermodynamic coupling effects, particularly between pressure and temperature, governing solubility dynamics. ## Toxicological potential of oil-based lubricants marketed in the EU - based on QSAR prediction, and harmonised classification. - Source: Toxicology (journals) - Date: 2026-09-19T00:00:00Z - Categories: Property Prediction, ADMET & Safety - Authors: J. Sørli, K. B. Frydendall, H. Shin, A. Sharma, N. Hadrup - Journal: Toxicology - DOI: 10.1016/j.tox.2026.154592 - External ID: 46537bd0f4b8a0b93e25cf5079a3fd437b4c7535 - Keywords: QSAR - Source URL: - Dashboard article: Abstract: Occupational exposure to oil-based lubricants involves inhalation, skin and mucosal surface contact. For hazard characterisation and safe-by-design, identification of toxicity endpoints of lubricant ingredients is desirable. To address this, we a) built a database of 178 oil-based lubricants available in the EU, by extracting CAS Registry numbers of 152 ingredients from safety data sheets (SDSs) (Product-DB), and b) compiled a database of 1032 lubricant ingredients from the Danish Product Registry (Registry-DB). We screened both databases for toxicity using QSAR predictions, EU harmonised classifications, IARC classifications and asthma databases. A dermal toxicity potential was observed for 14 and 17% of the substances in the Product- and Registry-DB, respectively. Reproductive toxicity potential was observed for 11 and 19% of the substances in the Product- and Registry-DB, respectively (7 and 13% when a low-specificity QSAR model is excluded). Although a carcinogenic potential was observed for 7 to 14% of the substances, none of the SDSs reported the mineral oil base stock as carcinogenic (one SDS flagged titanium dioxide). Other endpoints included hepatotoxicity (13 to 16% of the substances), and asthma (6% of the substances). Substances that could be considered high priority, based on the number of endpoints and products, identified across both databases, include N‑phenyl‑1‑naphthylamine, 2,6-di-tert-butyl-p-cresol (butylated hydroxytoluene), toluene, methyl methacrylate and maleic anhydride. QSAR prediction in particular pointed to substances with asthma, dermal and reproductive potential not identified by other screening methods, and can help prioritise substances for consideration in hazard characterisation and safe-by-design of oil-based lubricants. ## Treatment of an Advanced NSCLC Patient with a Rare OSBPL9-ALK Fusion: A Case Report - Source: Healthcare (journals) - Date: 2026-09-19T00:00:00Z - Authors: Liang Chen, Hu Chen, Lei Feng, Zi-Hao Zhang, Zhe-Peng Liu - Journal: Healthcare - DOI: 10.3390/healthcare14183085 - External ID: 10fa700d770160fc2d9259323650b411607fe2bc - Keywords: Autodock - Source URL: - Dashboard article: Abstract: ALK rearrangements define an important molecular subtype of lung adenocarcinoma, and the rearranged ALK serves as a well-established oncogenic driver in this malignancy. The most common type of mutation is the EML4-ALK fusion. Meanwhile, rare fusions are also frequently discovered with regard to ALK. However, clinical efficacy varies substantially across rare ALK fusion variants due to their distinct biological features, and individualized drug selection guided by fusion architecture and tumor characteristics is therefore required. This case presents the entire treatment process of a female patient with advanced lung adenocarcinoma who had an OSBPL9-ALK fusion. The patient was diagnosed in early 2021; by referring to previous reports and model prediction with AlphaFold 2 and Autodock Pymol, the new second-generation ALK inhibitor Ensartinib was chosen. After 24 months, due to disease progression, she switched to the third-generation ALK inhibitor, Lorlatinib, after model prediction. After 15 months, for further disease progression, the anti-angiogenic drug, anlotinib, was added to the treatment plan. Unfortunately, 6 months later, the disease progressed further, and the patient chose supportive treatment for economic reason. Nine months later, the patient died. The presentation of this case will provide a reference guideline for the subsequent treatment of lung adenocarcinoma with the same fusion. Meanwhile, model-based prediction for ALK fusion targeted therapy can provide a predictive reference for drug selection in patients presenting similar ALK rearrangements in the future. ## Unraveling the cytoprotective mechanism of polysaccharides in heat-stressed rabbits: antioxidant, anti-inflammatory, physiological status, and molecular docking assessment - Source: Veterinary Research Communications (journals) - Date: 2026-09-19T00:00:00Z - Categories: Docking & Screening - Authors: Mihaira H. Haddad, M. Bakeer, Mesharry M. M. Alharbi, Hanan M. Alharbi, K. Alwutayd, M. Shukry, S. Abdelnour, Mahmoud S. Abd-Allah - Journal: Veterinary Research Communications - DOI: 10.1007/s11259-026-11528-2 - External ID: 414de290fccb872b81d87c72b220ebe2d976428b - Keywords: molecular docking - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Virtual screening of Kv1.3 inhibitors and investigation of their molecular mechanisms in alleviating alcoholic liver injury in mice. - Source: European journal of medicinal chemistry (journals) - Date: 2026-09-19T00:00:00Z - Authors: Wen-Jun Zhen, Ji-Qing Ye, Guo-Qing Xia, Xi-Xi Chen, Yuan-Yuan Tian, Shi Chen, Meng-Qi Wang, Li-Xia Jia, Qi Zhang, Ye Wang, Jun Li, Jun-Chao Xu, Lei Zhang, Bao-Ming Wu - Journal: European journal of medicinal chemistry - DOI: 10.1016/j.ejmech.2026.119310 - External ID: 0797adc26c4a27e9350d54bba4098822e83c54a6 - Keywords: Virtual screening - Source URL: - Dashboard article: Abstract: Alcohol-related liver disease remains a global health challenge lacking effective pharmacotherapies. Given the role of the voltage-gated potassium channel Kv1.3 in immune regulation, this study aimed to identify potent Kv1.3 inhibitors for alcohol-related liver disease treatment. Guided by the GTEx database analysis linking Kv1.3 to hepatic inflammation, we conducted computer-aided virtual screening of 38,752 compounds. The lead compound, Z299576254, was selected and validated using membrane patch-clamp technology, demonstrating high affinity and significant inhibition of Kv1.3 outward currents. In a mouse model of alcohol-related liver disease, intraperitoneal administration of Z299576254 significantly alleviated hepatic injury, lipid deposition, and inflammatory responses without inducing hepatorenal toxicity. Mechanistic investigations using RNA-seq in RAW264.7 cells revealed that Z299576254 suppresses macrophage inflammatory responses by modulating the PI3K/AKT signaling pathway. Consequently, Z299576254 was identified as a safe and potent Kv1.3 inhibitor, offering a promising therapeutic strategy and molecular insight for managing alcohol-related liver disease. ## cboamd: A Machine Learning Molecular Dynamics Framework for Vibrational Strong Coupling - Source: arXiv (preprints) - Date: 2026-09-18T17:16:51Z - Categories: ML Potentials - Authors: Yifan Li, Roberto Car, Johannes Flick - External ID: 2609.22022v1 - Keywords: MLIP, Molecular Dynamics - Source URL: - Dashboard article: - PDF: Abstract: Under vibrational strong coupling (VSC), molecular vibrations hybridize with an optical cavity mode to form polaritons, offering a route to modify chemical and material properties without external driving. In this work, we develop a machine-learning interatomic potential (MLIP) based framework to study VSC inside optical cavities. By using the cavity Born-Oppenheimer approximation and treating the photonic degrees of freedom as an effective electric field, we provide a framework that can describe VSC solely based on the electronic ground-state potential energy surfaces (PES), electronic dipole moment, and polarizability, all quantities obtained outside the cavity. We train PES, polarization, and polarizability models to drive the molecular dynamics (MD) simulations of systems under VSC. We demonstrate the approach for both a single CO$\_2$ molecule and liquid CO$\_2$, showing that the polarizability renormalizes the effective cavity resonance: at fixed cavity frequency this renormalization renders the Rabi splitting strongly asymmetric, while with renormalized cavity frequency the symmetric splitting is capped by polarizability screening. The collective Rabi splitting of the liquid is connected quantitatively to the single-molecule splitting involving the $\\sqrt\{N/3\}$ orientational enhancement and the local-field enhanced effective charges of the coupled vibration, while the equilibrium pair structure of the CO$\_2$ liquid remains unchanged. These simulations are the first MLIP-driven simulations of collective VSC in the condensed phase and open further pathways to the exploration of chemical effects under VSC. ## fix uvt and fix pimd/uvt: A Unified LAMMPS Framework for Constant-Potential Constant-Temperature Molecular Dynamics - Source: arXiv (preprints) - Date: 2026-09-18T15:53:14Z - Authors: Li Fu, Yifan Li, Shenzhen Xu - External ID: 2609.21935v1 - Keywords: machine learning potentials, Molecular Dynamics - Source URL: - Dashboard article: - PDF: - Code: Abstract: Accurate simulations of electrochemical interfaces require the simultaneous treatment of constant-potential conditions, nuclear quantum effects, and sufficient configurational sampling. Integrating these capabilities within a general and efficient molecular dynamics (MD) framework remains challenging. In this work, we implement fix uvt and fix pimd/uvt in LAMMPS for constant-potential classical MD and path integral molecular dynamics (PIMD), respectively. We organize the class hierarchy to reuse LAMMPS's existing Nosé--Hoover chain thermostat routines and share nuclear propagation routines across PIMD integrators. A common interface connects these integrators to models that provide the electron-number derivative of the potential energy. We present three examples with accompanying input commands to guide users through constant-potential classical MD, thermostatted PIMD, and constant-potential PIMD, covering analytical models, liquid water, and electrochemical interfaces described by machine learning potentials. This work provides practical tools and guidance for large-scale constant-potential simulations incorporating nuclear quantum effects. ## Orbital-Free Surrogate Functionals Yield Transferable Interatomic Potentials and Electron Densities - Source: arXiv (preprints) - Date: 2026-09-18T15:10:48Z - Categories: ML Potentials - Authors: Simon Wagner, Marc K. Ickler, Manuel V. Klockow, Fred A. Hamprecht, Roman Remme - External ID: 2609.21882v1 - Keywords: density functional theory, DFT - Source URL: - Dashboard article: - PDF: - Code: Abstract: Orbital-free density functional theory seeks to compute the energy of an electronic system directly from its electron density, avoiding one-electron wave functions and thereby offering a route to scalable electronic structure calculations. Machine-learned orbital-free density functionals have recently achieved promising results on small organic molecules, predicting energies with sub-millihartree accuracy. However, their convergence in density optimization remains sensitive to hyperparameter tuning and architectural choices. Here, we extend the recently introduced (weak) surrogate functional framework - designed to predict ground-state electron densities only - to also yield their energy, resulting in "strong" surrogate functionals. We find that these learned functionals enable stable convergence across all tested neural network backbones, reducing electron density errors relative to the Kohn-Sham reference by an order of magnitude compared to previous OF-DFT methods. More importantly, the predicted energies are competitive with state-of-the-art machine-learned interatomic potentials (MLIPs) trained only on energies, while exhibiting superior generalization to larger, unseen systems. ## Enriching molecular Raman spectroscopy with vibrational strong coupling - Source: arXiv (preprints) - Date: 2026-09-18T09:32:24Z - Categories: Cheminformatics, Property Prediction, Spectra & Analytical - Authors: Matteo Castagnola, Anne Todsen Hansen, Morten Hanefeld Dziegiel, Anders Kristensen, Søren Raza, Simone Latini - External ID: 2609.21542v1 - Keywords: molecular fingerprints - Source URL: - Dashboard article: - PDF: Abstract: Raman spectroscopy is widely used for molecular identification in biological samples, but spectral congestion often obscures key molecular fingerprints. Strategies to enrich vibrational spectra with additional controllable features are therefore desirable. We theoretically show that vibrational strong coupling (VSC) can reshape Raman spectra by reorganizing vibrational energies and intensities. Near-resonant coupling enables the resolution of quasi-degenerate vibrational modes and redistributes Raman activity among molecular vibrations, brightening otherwise Raman-inactive modes. Off-resonant coupling mediates effective interactions between vibrations, leading to tunable intensity reorganizations and spectral shifts via cavity-controlled vibrational mixing. Using a microscopic theory of Raman scattering, we show that observing Raman signals from collective VSC polaritons requires appropriate geometries for the scattering setup, the sample, and the cavity environment, helping to explain why polaritonic Raman signatures have remained challenging to observe experimentally. ## SpecOpt: Contact-Diff Reasoning for Agentic Molecule Optimization Toward Binding Specificity - Source: arXiv (preprints) - Date: 2026-09-18T00:16:35Z - Categories: Cheminformatics, ADMET & Safety, LLMs & Agents - Authors: Thao Nguyen, Heng Ji - External ID: 2609.21165v1 - Keywords: ChEMBL, ADMET - Source URL: - Dashboard article: - PDF: Abstract: Off-target protein binding is a major source of adverse effects for small-molecule drugs, yet most structure-based molecular design methods focus on generating selective compounds de novo rather than improving the selectivity of existing, well- characterized drugs. We introduce specificity optimization (SpecOpt), a molecular design task that seeks constrained structural modifications to an existing compound that increase its binding preference for an intended target over known off-targets while preserving its structural identity and drug-like properties. To enable systematic evaluation, we construct a ChEMBL-derived benchmark from compound-target interaction data, identifying intended targets through curated drug-mechanism annotations and off- targets through measured activities. We then develop an agentic framework that docks each compound against its intended target and off-targets, compares the resulting poses through residue-aware atom-protein contacts, and provides these differential interactions to a large language model to propose targeted structural modifications. Candidates are retained only if they satisfy molecular similarity, ADMET, and target-off-target docking selectivity criteria. On 915 compounds, the agent improves the target- off-target binding gap for 84.8% of compounds, shifting the mean gap from -0.72 to +0.47 kcal/mol while maintaining a mean Tanimoto similarity of 0.72 to the starting compounds. Ablation studies identify residue-specific contact information as the critical optimization signal: replacing residue identities with binary contact indicators eliminates improvement on all 29 ablation compounds. These results establish SpecOpt as a distinct molecular design problem and demonstrate residue-aware differential interactions as an effective signal for improving the specificity of existing compounds. ## A critical review of advances in diffuse reflectance spectroscopy for rapid soil fertility assessment - Source: Discover Soil (journals) - Date: 2026-09-18T00:00:00Z - Authors: Tushar Kumar, K. Yadav, Ram Hari Meena, S. S. Lakhawat, G. Jat - Journal: Discover Soil - DOI: 10.1007/s44378-026-00314-w - External ID: acbe8329064ff0357216ef227b84218699914c6d - Keywords: convolutional neural - Source URL: - Dashboard article: Abstract: The growing pressure on soil resources from climate change, land degradation, and the increasing demand for food security has intensified the need for rapid, accurate, and cost-effective methods of soil characterisation. Diffuse reflectance spectroscopy (DRS), operating across the visible-near infrared (Vis-NIR: 350–2500 nm) and mid-infrared (MIR: 4000–400 cm⁻¹) regions of the electromagnetic spectrum, has emerged as a powerful analytical platform that addresses these challenges. By exploiting the diagnostic absorption features arising from molecular vibrations and electronic transitions of key soil chromophores (organic matter, clay minerals, iron oxides, calcium carbonates, and water), DRS enables the simultaneous estimation of numerous soil properties from a single spectral scan, without generating chemical waste and with minimal sample preparation. This review synthesises more than nine decades of scientific progress in soil spectroscopy, from the early spectral libraries of the 1930s to the contemporary integration of machine-learning and deep-learning frameworks. We examine the mechanisms underlying spectral absorption, critically assess the comparative performance of Vis-NIR and MIR techniques, evaluate pre-processing strategies from Savitzky-Golay smoothing and standard normal variate correction to derivative transformations, and benchmark the full spectrum of prediction models, from partial least squares regression (PLSR) and multivariate adaptive regression splines (MARS) to support vector regression (SVR), random forests, and convolutional neural networks. The review also critically documents performance across a comprehensive range of soil properties including texture, organic carbon, cation exchange capacity, pH, electrical conductivity, and macro- and micronutrients, highlighting consistent achievements as well as persistent limitations. A central argument of this review is that while the field has achieved remarkable predictive capabilities for certain core properties, key assumptions regarding model transferability, chromophore linearity, and pre-processing universality remain empirically untested. The paper concludes by identifying the five most consequential unanswered research questions in the discipline, with particular attention to the challenges posed by the heterogeneous agro-ecological landscapes of India. Recommendations for a methodologically rigorous path forward are provided. To make the comparison more transparent, selected validation results are aggregated by fertility property rather than pooled across incompatible studies. The comparison supports selective use of DRS for clay or texture, organic matter, cation-exchange capacity, and total nitrogen, while electrical conductivity, available phosphorus, potassium, and DTPA-extractable micronutrients remain dependent on the calibration domain and validation design. ## A hybrid BiLSTM transformer model for drug synergy prediction - Source: Discover Artificial Intelligence (journals) - Date: 2026-09-18T00:00:00Z - Categories: Cheminformatics - Authors: Sahar Abbasi Rostami, A. Lakizadeh - Journal: Discover Artificial Intelligence - DOI: 10.1007/s44163-026-02262-4 - External ID: 18109a1239de59d8779c2b20ae5ed83b9e56d228 - Keywords: SELFIES, LSTM, transformer, self attention, molecular representations - Source URL: - Dashboard article: Abstract: Accurately predicting synergistic drug combinations remains a critical challenge in computational pharmacology, with important implications for the development of combination therapies in oncology. However, in vitro screening of drug combinations for synergy is both time-consuming and labor-intensive, as the number of possible combinations grows exponentially. Although several computational methods have been proposed to predict synergistic drug pairs, more effective modeling of the complex multidimensional relationships among drug compounds is still needed. In this study, we present BT-Synergy, a task-specific hybrid deep learning model for drug synergy prediction. The model integrates BiLSTM modules within Transformer blocks to capture complementary sequential and contextual patterns from SELFIES-based molecular representations. In this design, bidirectional LSTM layers complement self-attention with sequential inductive biases, enabling the model to capture local molecular patterns as well as long-range contextual dependencies. In parallel, protein features are extracted using pre-trained protein language models and aggregated to form biologically informed cell-line representations. These integrated embeddings, which combine drug, protein, and cell-line information, are then processed by a multi-layer perceptron to predict drug synergy. Experimental results on benchmark datasets show that BT-Synergy effectively predicts synergistic interactions, achieving an accuracy of 0.8458. These findings highlight the effectiveness of the proposed architecture in capturing complex biochemical interactions and improving predictive performance in drug synergy prediction. ## A Molecular Generation–Validation Workflow Combining Reversible Junction Tree Reinforcement Learning and Semi-empirical Quantum Chemistry: A Case Study on Molecular Solar Thermal Fuels - Source: ChemRxiv (preprints) - Date: 2026-09-18T00:00:00Z - Categories: Design de novo - Authors: Qingyu Zhu, Ryan Lingg, Reuben Szabo, Tim Kowalczyk, Daisuke Yokogawa - DOI: 10.26434/chemrxiv.15009066/v1 - External ID: 10.26434/chemrxiv.15009066/v1 - Keywords: Molecular Generation, generative models, Reinforcement Learning, RL, explore chemical space, Quantum Chemistry - Source URL: - Dashboard article: Abstract: Molecular generative models can rapidly explore chemical space, but conventional metrics alone do not establish whether proposed structures possess application-relevant physical properties. To address this limitation, we developed an integrated generationvalidation framework coupling Reversible Junction Tree Reinforcement Learning (RJT- RL) with automated density-functional tight-binding (DFTB) evaluation. Azobenzenebased molecular solar thermal fuels served as a representative case study. RJT-RL was guided by maximum Tanimoto similarity to a reference dataset, while reference and generated molecules were evaluated under identical DFTB calculation conditions to obtain the energy difference between the cis and trans isomers (∆ E iso ) and the S 0 → S 1 excitation energy (∆ E ex ). As training progressed, the ∆ E iso distribution extended toward higher values, while ∆ E ex values were more concentrated within the visible-light window, yielding candidates within the selected joint screening region, although neither descriptor was included in the reward. These results show that automated semi-empirical post-generation validation can identify candidates with targeted property combinations and reveal information beyond conventional molecular generation metrics. ## AI Prediction of Future Antibiotics: Genome-Predicted Antimicrobial Peptides Are Stained with the Amino Acid Signature of Man-Made Synthetic Peptides - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-18T00:00:00+00:00 - Categories: Cheminformatics - Authors: Guangshun Wang - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c02831 - Source URL: - Dashboard article: Abstract: This letter compares the amino acid signatures of natural and artificial intelligence (AI)-predicted antimicrobial peptides. While natural peptides from different sources possess distinct signatures, the amino acid compositions of peptides predicted from venoms, amphibians, insects, and bacteria shared a common feature with higher leucine, lysine, and arginine similar to those of man-made synthetic peptides. This stunning finding underscores the need of the right data set for improving future AI prediction of genomic antimicrobial peptides. ## An intelligent OCR-LLM prescription review system for inpatient narcotic and psychotropic drugs - Source: Medicine (journals) - Date: 2026-09-18T00:00:00Z - Categories: LLMs & Agents - Authors: Rong-Wu Ma, Kai Xu, Ming-Ming Zhang, Xiang-Peng Li, Xing-Rong Ma, Jun-Yu Wang, Jing Li - Journal: Medicine - DOI: 10.1097/MD.0000000000050741 - External ID: d957f60f5b5760f634f1a574e8caac236aa39003 - Keywords: LLM - Source URL: - Dashboard article: Abstract: This study aimed to construct an intelligent prescription review system for inpatient narcotic and psychotropic drugs based on deep learning-based optical character recognition (OCR) and a local large language model (LLM) and evaluate its review performance and closed-loop management value. This study adopted a combined system development and retrospective validation design. Prescription images were first processed using OCR for text recognition and layout reconstruction. A local LLM was then applied to convert unstructured prescription text into standardized JavaScript Object Notation-formatted data. Subsequently, a locally developed rule base for narcotic and psychotropic drugs was used to evaluate prescription completeness, diagnosis-based dosage limits, medication instructions, physician signatures, and residual drug handling for injectable formulations. Using pharmacists’ manual review results as the reference standard, 500 inpatient prescriptions for narcotic and psychotropic drugs were retrospectively included. System performance was evaluated in terms of accuracy, sensitivity, specificity, positive predictive value, negative predictive value, F1 score, and Cohen’s kappa coefficient. The system successfully implemented prescription image recognition, text structuring, rule-based validation, signature verification, residual drug calculation, and triggers for manual pharmacist review. In the retrospective validation, compared with pharmacists’ manual review, the system identified 144 true positives, 16 false positives, 334 true negatives, and 6 false negatives. The overall accuracy was 95.6%, sensitivity was 96.0%, specificity was 95.4%, positive predictive value was 90.0%, negative predictive value was 98.2%, F1 score was 0.929, and Cohen’s kappa coefficient was 0.897. The proposed OCR-LLM-based intelligent prescription review system for inpatient narcotic and psychotropic drugs demonstrates high accuracy and strong consistency. It can serve as a front-end screening and decision-support tool for pharmacists and provides technical support for standardized, traceable, and closed-loop management of specially controlled medications. ## Analysis of structural errors in the AlphaFold DB v4 and v6 - Source: Journal of Cheminformatics (journals) - Date: 2026-09-18T00:00:00+00:00 - Categories: Targets & Structures - Authors: Lukáš Bohuš, Tomáš Svoboda, Ondřej Schindler - Journal: Journal of Cheminformatics - DOI: 10.1186/s13321-026-01285-4 - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Artificial Intelligence-Assisted Dopant Discovery toward Air-Stable Sulfide Solid-State Electrolytes - Source: ChemRxiv (preprints) - Date: 2026-09-18T00:00:00Z - Categories: LLMs & Agents - Authors: Zhuo Chen, Lin Hong, Xi Zhang - DOI: 10.26434/chemrxiv.15009059/v1 - External ID: 10.26434/chemrxiv.15009059/v1 - Keywords: Gibbs free energy - Source URL: - Dashboard article: Abstract: Sulfide solid-state electrolytes (SSEs) are highly promising candidates for all-solid-state batteries (ASSBs), but the severe degradation triggered by moisture exposure remains a major obstacle to their large-scale application. Despite compositional doping could effectively enhance the air stability, there is currently no relevant screening strategy to support the efficient exploration of dopants. Herein, we develop an artificial intelligence-assisted dopant screening platform for air-stable sulfide SSEs (LPSC). Retrieval-augmented system combined with large language model could effectively prescreen the dopants that enhance both the ionic conductivity, and interfacial compatibility of the electrolyte. More importantly, the impact of dopants on the air stability of electrolytes was quantified by the minimum Gibbs free energy change () of the hydrolysis reaction, which could be predicted by a composition-based machine learning model. Guided by this strategy, several promising dopants, including BiF3, CoF3, InF3, and SnF4 were successfully identified. As a proof of concept, LPSC-BiF3 electrolyte demonstrates superior air stability, delivering a high ionic conductivity retention of 91% after exposure to air with 10% RH for 6 h. Consequently, the exposed LPSC-BiF3 electrolyte enabled excellent cycling stability in Li-In||NCM811 full cells. The proposed screening strategy significantly promotes the exploration of high-performance dopants for air-stable sulfide SSEs. ## Assessing the Generalizability of Machine Learning and Physics-Based Methods with DNA-Encoded Libraries - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-18T00:00:00+00:00 - Categories: Docking & Screening, Targets & Structures, Library Design - Authors: Marissa Dolorfino, Daniel Santos Perez, Yao Fu, Shu-Hang Lin, Sean McCarty, Matthew J. O’Meara, Terra Sztain - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c01611 - Keywords: Boltz 2, DNA Encoded Libraries - Source URL: - Dashboard article: Abstract: Predicting protein–ligand binding is a central challenge in computational drug discovery, and while machine learning (ML) and cofolding methods have advanced rapidly, their ability to generalize beyond training or parametrization regimes remains insufficiently understood. DNA-encoded libraries (DELs) enable ultralarge screening of billions of molecules simultaneously, providing a useful testbed for evaluating these approaches at scale. A recent NeurIPS competition revealed that even top-performing ML models trained on DEL data failed at generalizing to out-of-distribution (OOD) chemical space. We investigated whether integrating structural modeling could bridge this generalization gap. We systematically assessed state-of-the-art ML, docking, and cofolding methods, including Schrödinger Glide, Rosetta GALigandDock, and Boltz-2 with three biologically diverse protein targets screened against libraries containing multiple DEL synthesis formats. While ML excels in-distribution, OOD hit discrimination is dependent on both the target and ligand context, with no single method consistently dominating. These findings demonstrate that benchmark performance alone is insufficient to predict OOD performance, highlighting the need for system-dependent evaluation of binding prediction methods. We provide an open-source package for assessing protein–ligand prediction methods and analyzing high-throughput screening data: DEL-iver. ## Bardoxolone methyl exerts an inhibitory effect on respiratory syncytial virus infection by downregulating IL-6 via the IKKβ/NF-κB signaling pathway - Source: Microbiology Spectrum (journals) - Date: 2026-09-18T00:00:00Z - Categories: Docking & Screening - Authors: Xun-Ping Wu, Hongping Wang, Ling Zhang, Bing-Yao Wang, Xin-Ran Tan, Zhu Li, J. Nie, Dai-Shun Liu - Journal: Microbiology Spectrum - DOI: 10.1128/spectrum.03825-25 - External ID: 5245314034bf6eeef0321cb432dc55eff867cec0 - Keywords: molecular docking - Source URL: - Dashboard article: Abstract: Respiratory syncytial virus (RSV) causes severe lower respiratory tract infections in infants, older adults, and immunocompromised individuals, but therapeutic options for active infection remain limited. Interleukin-6 (IL-6) contributes to RSV replication and RSV-induced inflammation, and the IKKβ/NF-κB pathway is a major regulator of IL-6 expression. Bardoxolone methyl (BXM) was identified by network medicine as a potential anti-RSV compound, with molecular docking suggesting interactions with the IKKβ/NF-κB pathway. We investigated whether BXM inhibits RSV infection through modulation of the IKKβ/NF-κB/IL-6 axis in RSV-infected BEAS-2B cells and BALB/c mice. RSV infection increased IL-6 expression and activated IKKβ/NF-κB signaling, whereas IL-6 knockdown reduced viral replication and inflammatory mediator expression. BXM significantly suppressed RSV replication, as shown by RT-qPCR, TCID₅₀, plaque reduction assays, and Western blotting, and decreased IL-6 expression in vitro. Mechanistically, BXM attenuated RSV-induced phosphorylation of IKKβ and NF-κB. The IKKβ inhibitor TPCA-1 produced similar inhibitory effects, whereas recombinant human IL-6 add-back partially restored inflammatory mediator expression in BXM-treated RSV-infected cells. In mice, BXM reduced lung viral titers, RSV-F expression, IL-6 expression, IKKβ/NF-κB activation, and RSV-induced lung injury. These findings suggest that BXM suppresses RSV replication and inflammation, at least in part, by modulating the IKKβ/NF-κB/IL-6 axis, supporting its potential as a therapeutic candidate for RSV infection. IMPORTANCE Respiratory syncytial virus (RSV) is a major cause of respiratory illness, especially in infants, older adults, and people with weakened immune systems. Although preventive strategies are available for some high-risk groups, treatment options for active RSV infection remain limited. This study identifies bardoxolone methyl (BXM) as a potential anti-RSV compound and shows that it can reduce both viral replication and virus-induced inflammation in cell and mouse models. Importantly, our findings suggest that BXM acts, at least in part, by reducing IL-6-related inflammatory signaling, which is closely associated with RSV disease severity. These results provide experimental evidence supporting BXM as a promising candidate for further development against RSV infection and offer insight into a potential host-targeted therapeutic strategy. Respiratory syncytial virus (RSV) is a major cause of respiratory illness, especially in infants, older adults, and people with weakened immune systems. Although preventive strategies are available for some high-risk groups, treatment options for active RSV infection remain limited. This study identifies bardoxolone methyl (BXM) as a potential anti-RSV compound and shows that it can reduce both viral replication and virus-induced inflammation in cell and mouse models. Importantly, our findings suggest that BXM acts, at least in part, by reducing IL-6-related inflammatory signaling, which is closely associated with RSV disease severity. These results provide experimental evidence supporting BXM as a promising candidate for further development against RSV infection and offer insight into a potential host-targeted therapeutic strategy. ## Computational drug repurposing against histone acetyltransferase p300 for non-hormonal treatment of endometriosis using virtual screening and molecular dynamics simulations - Source: Scientific Reports (journals) - Date: 2026-09-18T00:00:00+00:00 - Authors: Hajar Erraji, Mohammed Hakmi, Mohammed Zarqaoui, Noureddine Louanjli, Souad Kartti, Bouchra Ghazi - Journal: Scientific Reports - DOI: 10.1038/s41598-026-68337-5 - Keywords: virtual screening, molecular docking, molecular dynamics - Source URL: - Dashboard article: Abstract: Endometriosis is a gynecological disorder, characterized by debilitating pain and commonly associated with infertility. Medical therapies are often ineffective, with 25–34% of patients exhibit no or poor response, and many experiencing intolerable side effects that limit long term use. P300 is a critical histone acetyltransferase that promotes inflammation and invasion through epigenetic modulation of immune cells, and upregulation of pro-inflammatory gene expression. In this study, we aimed to identify potential inhibitors of p300 for treating endometriosis using computational drug repurposing. We performed a molecular docking-based virtual screening of 2,083 Food and Drug Administration (FDA) approved drugs against the crystal structure of p300 ligand-binding domain, followed by molecular dynamics (MD) simulations and Molecular Mechanics Poisson Boltzmann Surface Area (MM-PBSA) analysis to further assess the stability and the binding interactions of the top-ranked drugs. Three compounds, Naldemedine, Bictegravir and Zavegepant, exhibited favorable docking affinities and interaction profiles. Further evaluation using 300 ns molecular dynamics simulations and MM-PBSA analyses demonstrated that Bictegravir exhibited the most favorable binding in complex with p300. Lastly, these findings underscore the potential of bioinformatics approaches in providing valuable resources and novel perspective on therapeutic strategies that are often overlooked in the context of endometriosis. ## Computational Performance of Programming Languages in Mathematical Biology: A Ten-Language Evaluation Across Six Modeling Regimes - Source: Symmetry (journals) - Date: 2026-09-18T00:00:00Z - Authors: Yi Zheng, Qiu-Ming Luo - Journal: Symmetry - DOI: 10.3390/sym18091559 - External ID: bd0c540477d1f64eb2fca73ee95b8a89aa66cf82 - Keywords: diffusion models - Source URL: - Dashboard article: Abstract: Mathematical biology spans multiple computational regimes—from ordinary differential equation models of gene regulatory networks to stochastic simulation of chemical kinetics and reaction-diffusion models of spatial pattern formation—each imposing distinct computational demands on the software infrastructure that executes them. The choice of programming language for implementation of these mathematical models carries quantitative performance consequences that have not been systematically measured across the range of models employed in contemporary mathematical biology. This study provides a computational performance evaluation across ten languages (Python, Julia, Rust, C, C++, C\#, F\#, Go, Java, and R) and six mathematical biology modeling regimes: deterministic ODE integration, stochastic chemical kinetics, parameter-space exploration, reaction-diffusion spatial modeling, agent-based discrete simulation, and Bayesian parameter inference. Execution time, peak memory footprint, cyclomatic complexity, type-conversion density, and the semantic alignment between data structures and biological state representation were recorded under a uniform experimental protocol. Under a unified hand-coded Dormand–Prince 5(4) integrator, native implementations outperform managed-runtime counterparts by nearly two orders of magnitude for ODE integration, a differential that shrinks sharply once library-delegated solvers are removed from the comparison; the gap contracts below 55× when stochastic kinetics shift the bottleneck from arithmetic throughput to branch resolution. In memory-bandwidth-limited reaction-diffusion modeling, native and just-in-time implementations converge to within a few percent. Agent-based models expose a distinct regime-dependent overhead: immutable-by-default collection semantics impose disproportionate cost during mutation-intensive computation. Peak memory varies by roughly two orders of magnitude across languages, directly affecting deployment density for large-scale simulation. Code-structural measurements confirm that algorithmic form enforces a floor on cyclomatic complexity, irrespective of language, while type strictness and mutation semantics generate substantial differences in per-line cognitive load. These results provide a quantitative foundation for computational tool selection in mathematical biology, challenging universal language recommendations and supporting choices grounded in the algorithmic character of each modeling regime. ## CoTAR: Topology and Atomic State Reconstruction in Condensed Phases - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-18T00:00:00+00:00 - Categories: Cheminformatics, Property Prediction, ML Potentials - Authors: Hodaka Mori, Yu Miyazaki, Takechika Kikkawa - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c02184 - Keywords: graph neural network, GNN, molecular representations, molecular dynamics, force fields - Source URL: - Dashboard article: Abstract: Universal machine learning interatomic potentials (uMLIPs) enable performing condensed-phase molecular dynamics (MD) simulations with accuracy approaching that of first-principles; however, their lack of explicit molecular topology limits bond-aware analysis and reconnection to classical force fields. This study presents CoTAR, a hybrid graph neural network (GNN)–hidden Markov model (HMM) framework that reconstructs bond connectivity and order, formal charges, and unpaired electrons from atomic species and coordinates by combining learned local environments, chemical constraints, and temporal smoothing. After joint fine-tuning with 20 labeled snapshots per benchmark system, CoTAR achieved category-macro bond and conditional bond-order F1 scores of 0.992 and 0.998 and chemically valid snapshots of 82.1% across 128 uMLIP systems. Zero-shot tests were successful for all tested nonionic systems but revealed persistent formal-charge errors in ionic mixtures. For successfully reconnected systems, the reconstructed-topology simulation results showed close agreement with the reference-topology results for the density, element-resolved diffusion, and intermolecular radial distribution functions. These results establish CoTAR as a practical route from topology-free uMLIP trajectories to chemically complete molecular representations within the present scope of nonreactive, topology-preserving systems. ## Cross-phenotype integration of eQTL-based Mendelian randomization and transcriptomics prioritizes BMP2K in IgA vasculitis with nephritis - Source: Frontiers in Immunology (journals) - Date: 2026-09-18T00:00:00Z - Categories: Docking & Screening - Authors: Su Peng, Chuan Xie, Chu-Yun Qian, Ping Yuan, Shi-You Guan, Xiang-Mei Zhang, Yan Chen, Pei Huang, Zuo-Chen Du - Journal: Frontiers in Immunology - DOI: 10.3389/fimmu.2026.1757142 - External ID: 89a8eeb57473a1586c0730437791dfd81f52611c - Keywords: molecular docking - Source URL: - Dashboard article: Abstract: To prioritize candidate genes and compounds potentially associated with IgA vasculitis with nephritis (IgAVN) by integrating genetic and transcriptomic evidence across IgAVN, IgA vasculitis (IgAV), and IgA nephropathy (IgAN), followed by preliminary cellular and pharmacological validation. Expression quantitative trait locus (eQTL)-based two-sample Mendelian randomization (MR) was performed using FinnGen summary statistics for IgAVN, IgAV, and IgAN. MR-prioritized genes were integrated with differentially expressed genes from the IgAVN transcriptomic dataset GSE102114. Fisher’s method was used to combine convergent evidence, and exploratory immune cell-specific single-cell eQTL-based MR was conducted to identify potential cell-type-relevant signals. Candidate compounds were screened using drug-signature analysis and evaluated by molecular docking. BMP2K mRNA expression was measured by RT-qPCR in peripheral blood mononuclear cells (PBMCs) from children with IgAVN ( n = 13), children with IgAV ( n = 10), and healthy controls ( n = 13). Functional and pharmacological effects were evaluated in lipopolysaccharide (LPS)-stimulated THP-1 cells using siRNA-mediated knockdown, CCK-8, RT-qPCR, Western blotting, and ELISA. MR analysis identified 124 genes with nominally significant risk-increasing associations with IgAVN, 132 with IgAV, and 119 with IgAN. Integration with IgAVN transcriptomic data prioritized four candidate genes: BMP2K , GYPB , IGFBP7 , and MISP3 . Among these, BMP2K was the only gene supported across all three IgA-related phenotypes and the IgAVN transcriptomic dataset. BMP2K mRNA expression was significantly higher in PBMCs from children with IgAVN than in healthy controls. Exploratory single-cell eQTL-based MR suggested a nominal monocyte-related association between genetically predicted BMP2K expression and IgAVN risk. In THP-1 cells, BMP2K knockdown reduced LPS-induced secretion of IL-6, IL-8, and TNF-α. Drug-signature analysis and molecular docking nominated ruboxistaurin as a candidate compound. Ruboxistaurin attenuated LPS-induced inflammatory cytokine expression and secretion and was associated with a statistically significant reduction in BMP2K expression. Integrative genetic, transcriptomic, and preliminary experimental evidence prioritized BMP2K as a candidate gene for IgAVN. Ruboxistaurin showed anti-inflammatory activity and was associated with lower BMP2K protein abundance in vitro ; however, direct target engagement and BMP2K-specific activity were not established. Larger independent genetic datasets, colocalization analyses, renal tissue validation, and disease-relevant mechanistic models are required. ## De Novo Design and Structural Optimization of Mn(salen)‐Based Artificial Metalloenzymes for Asymmetric Sulfoxidation - Source: Angewandte Chemie International Edition (journals) - Date: 2026-09-18T00:00:00+00:00 - Authors: Jing‐Xiang Wang, Yunling Deng, Indrek Kalvet, Amira Haque, Huiguang Dai, David Baker, Yi Lu - Journal: Angewandte Chemie International Edition - DOI: 10.1002/anie.5852828 - Keywords: De Novo Design - Source URL: - Dashboard article: Abstract: Artificial metalloenzymes (ArMs) exhibit exceptional selectivity, yet extending their reactivity beyond native cofactors remains a major challenge. While previous designs using native protein scaffolds to incorporate nonnative cofactors have been reported, de novo protein design enables tailored scaffolds that incorporate nonnative cofactors, unlocking transformations inaccessible to natural enzymes. Here, we report the computational design of de novo proteins that bind Mn(salen)‐based complexes for asymmetric sulfoxidation. The resulting ArMs outperform the free cofactor, achieving up to 45% yield and an enantiomeric ratio (e.r.) of 26:74 under optimized conditions. A 1.5 Å resolution crystal structure confirms the designed architecture and reveals key secondary‐sphere interactions that govern reactivity. Guided by these insights, rational mutagenesis enhanced performance up to 79% yield and an e.r. up to 16:84. This work establishes a general strategy for integrating complex nonnative cofactors into de novo scaffolds, enabling selective catalysts for reactions beyond the reach of natural enzymes. ## Design, synthesis, in silico studies, and biological evaluation of tosyl-substituted thiazoles, thiazolidin-4-ones, and chromenes as potential anticancer agents - Source: RSC Advances (journals) - Date: 2026-09-18T00:00:00Z - Categories: Docking & Screening - Authors: Gehad E. Said, Sonia Samy, E. Abdel-Galil, Hatem E. Gaffer, E. Abdel-Latif - Journal: RSC Advances - DOI: 10.1039/d6ra06404c - External ID: 8e2415079fd703ff39ae089918b131a1f984ef3b - Keywords: molecular docking, pharmacokinetic, receptor, DFT, IC50 - Source URL: - Dashboard article: Abstract: A novel series of thiazole, thiazolidin-4-one, and chromene derivatives incorporating a 4-methylbenzenesulfonate (tosylate) moiety were designed, synthesized, and evaluated for their anticancer potential. All structures were elucidated using IR, 1H NMR, 13C NMR, and MS spectroscopy. Anticancer potential was assessed by MTT assay against HepG-2 (hepatocellular carcinoma) and MCF-7 (breast cancer) cell lines, with doxorubicin as a reference. Against MCF-7, compounds 3d (IC50 = 28.2 ± 2.08 µM), 6a (30.0 ± 0.82 µM), and 8a (25.9 ± 1.35 µM) exhibited relatively enhanced cytotoxicity. For HepG-2, derivatives 3a (46.5 ± 1.23 µM), 3e (49.1 ± 0.95 µM), and 3d (54.6 ± 1.34 µM) showed moderate cytotoxic activity relative to doxorubicin. Also, all the synthesized compounds were subjected to molecular docking against the receptor (PDB ID: 1X7B) when compared with the standard drug doxorubicin. However, 3b had the best binding energy (6.4543 kcal mol−1), resulting in a π–cation interaction with phenyl and Lys471, and was closely followed by 3c (6.3568 kcal mol−1), 8c (6.3141 kcal mol−1), and 8b (6.3020 kcal mol−1). From DFT analysis on the most active molecules 3d, 6a, and 8a, it is clear that molecule 8a exhibits the lowest energy difference between its HOMO and LUMO orbitals (Egap = 3.39 eV) as well as highest molecular softness value (δ = 0.59), which may contribute to its enhanced chemical reactivity and observed biological activity compared to other molecules. Moreover, the pharmacokinetic characteristics of the thirteen synthesized analogues using SwissADME exhibited poor solubility, possessed low GI absorption, and unveiled no BBB permeability. The newly developed analogs possessed a high lipophilicity, larger polar surface area and poor predicted gastrointestinal absorption, mainly due to the presence of the sulfonamide/sulfonate-functionalized core structure. In addition to this, some of the compounds exhibited an inhibitory effect on some of the important CYP450 enzymes (CYP2C19, CYP2C9 and/or CYP3A4). The pharmacokinetic challenges pose a great need to optimize the structures before preclinical evaluation. Overall, this study provides integrated experimental and computational insights into the anticancer potential of tosylate-bearing heterocyclic scaffolds, addressing gaps in understanding their structure–activity relationships. The identified promising compounds may serve as useful starting points for further structural optimization and mechanistic studies toward the development of improved anticancer candidates. ## Developing SCL2205 : A Protein Sequence-based Spatial Modelling Dataset for the Protein Language Model Frontier - Source: Bioinformatics (journals) - Date: 2026-09-18T00:00:00+00:00 - Authors: Daniel Ouso, Gianluca Pollastri - Journal: Bioinformatics - DOI: 10.1093/bioinformatics/btag666 - Source URL: - Dashboard article: - Code: Abstract: Motivation Deep learning (DL) has substantially advanced protein subcellular localisation (SCL) prediction, yet its potential remains constrained by suboptimal input preparation and limited high-quality reference data. Furthermore, existing state-of-the-art (SoTA) predictors suffer from performance metric inflation due to unmitigated training-to-testing data leakage during homology augmentation. We address these challenges by introducing SCL2205, a leak-minimised benchmark dataset and pipeline curated specifically to support trustworthy, scalable, and reproducible DL-based SCL modelling. Results SCL2205 was constructed from the universal protein knowledgebase (UniProtKB) using rigorous preprocessing, manual label mapping, and stringent partitioning. When evaluated on independent test sets, SCL2205 yielded up to a 10.8 percentage point improvement in macro area under the precision–recall curve (PR-AUC) over SoTA baselines (mean Δ 95% CI=0.07−0.12 ), with maximum benefits observed when paired with modern protein language models (PLMs). Crucially, we quantify for the first time a systemic 5.2%±0.32 data leakage rate in conventional homology augmentation workflows—even when restricting sequence similarity searches to just 10% of the training set. Availability and Implementation The dataset is openly available on Dryad under a CC0 1.0 Universal licence (https://doi.org/10.5061/dryad.2ngf1vj1t). The dataset interface is available as an installable Python package, p-scldata (v2026.2.0), under the MIT licence on the Python Package Index (PyPI). Full code and data repositories are hosted on GitHub (https://github.com/ousodaniel/scldata) and archived on Zenodo (https://doi.org/10.5281/zenodo.21796423). Supplementary Information Supplementary File S1 contains code snippets, per-class PR-AUC breakdowns, class prevalence details, statistical comparison tests, and supplementary figures. Supplementary File S2 contains the exact mapping used in curation. ## Discovery of Novel Meta-Diamide Derivatives with Improved Safety as GABA Receptor-Targeted Insecticides by the Scaffold-Hopping Strategy and 3D-QSAR Models - Source: Journal of Agricultural and Food Chemistry (journals) - Date: 2026-09-18T00:00:00Z - Categories: Property Prediction, Docking & Screening, ADMET & Safety - Authors: Hua-Nan Zeng, Rui-Ying Ma, Yu Li, Lei Zhou, Xiao-Yan Xu, Xin-Xin Xu, Zi-Wen Wang, Ming-Zhi Huang, Qingmin Wang - Journal: Journal of Agricultural and Food Chemistry - DOI: 10.1021/acs.jafc.6c12171 - External ID: bb7b3635ac56c5e043dbe7bf68c88c94347e0ab0 - Keywords: QSAR, Molecular docking, Receptor - Source URL: - Dashboard article: Abstract: The extensive application of meta-diamide insecticides has raised serious concerns regarding their toxicity to aquatic nontarget organisms, necessitating the development of safer alternatives. In this study, a combined in silico approach integrating a scaffold-hopping strategy and three-dimensional quantitative structure–activity relationship (3D-QSAR) models was employed to discover novel meta-diamide derivatives as GABA receptor-targeted insecticides with improved safety profiles. Reliable 3D-QSAR models were constructed based on meta-diamide derivatives developed through a scaffold-hopping strategy, which ultimately enabled the design and synthesis of the novel compound A31. The bioassay results showed that compound A31 displayed good insecticidal activity against Plutella xylostella, with an LC50 value of 0.13 mg/L, and it demonstrated greater safety for zebrafish. Molecular docking studies confirmed that compound A31 maintained favorable binding modes within the GABA receptor. These findings provide valuable insights into the rational design of environmentally compatible meta-diamide insecticides and offer promising candidates for further experimental validation. ## Discovery of Pyranochromone Derivatives as TLR4-MD2 Inhibitors for Alleviating Osteoarthritis - Source: Journal of Natural Products (journals) - Date: 2026-09-18T00:00:00Z - Categories: Docking & Screening - Authors: Ming-Shi Pan, Rong Wang, Jian-Qiang Qian, Wei Shi, Zi-Hao Wang, Yu-Hang Lian, Xiao-Qi Zhang, Wen-Cai Ye, Fei Xiong, Xiao-Long Hu, Hao Wang - Journal: Journal of Natural Products - DOI: 10.1021/acs.jnatprod.6c00794 - External ID: 57e1225c9bd03e082c59a81ff52b095a74599cec - Keywords: molecular docking, IC50 - Source URL: - Dashboard article: Abstract: Calomembranone G, isolated from Calophyllum membranaceum, displays in vivo anti-inflammatory activity in models of arthritis and acute lung injury. Using natural product calomembranone G as a hit compound and leveraging a strategy of structural simplification, a total of 66 pyranochromone derivatives were synthesized and their structure−activity relationships were analyzed. Among these analogues, compound PC-C7 displayed the most potent anti-inflammatory activity, inhibiting NO production with an IC50 value of 0.17 ± 0.07 μM. Surface plasmon resonance, molecular docking, and co-immunoprecipitation assays demonstrated that PC-C7 disrupts the formation of the TLR4−MD2 complex. In vivo, oral administration of PC-C7 (10 or 20 mg/kg) alleviated joint inflammation and chondrocyte catabolism, through suppressing TLR4-mediated inflammatory signaling. These findings demonstrated that PC-C7 might serve as a drug candidate for osteoarthritis therapy. ## Discovery of Pyrimidinedione Derivatives Containing Oxime Ester/Carbamate Fragments as Potent Protoporphyrinogen IX Oxidase Inhibitors - Source: Journal of Agricultural and Food Chemistry (journals) - Date: 2026-09-18T00:00:00Z - Categories: Docking & Screening, Free Energy & MD - Authors: Wei Zhang, Ke-Shi Luo, Di Li, Xian-Ying Tang, Xiu-Hai Gan - Journal: Journal of Agricultural and Food Chemistry - DOI: 10.1021/acs.jafc.6c01948 - External ID: 857f3e1165fc6d4e07829bcc97f8e8619c94a4c0 - Keywords: Molecular docking, binding free energy - Source URL: - Dashboard article: Abstract: To continue our research on developing PPO inhibitors, we replaced the triazinone core of compounds B, G, and H from our previous work with pyrimidinediones, employing a scaffold-hopping strategy to design and synthesize compounds I, J, and K. Compound I18 exhibited potent NtPPO inhibition (Ki = 10.9 nM) and broad-spectrum herbicidal activity comparable to saflufenacil, providing 100% control of 29 weeds at 75 g a.i./ha. It also showed improved crop safety, causing only 30% peanut injury and thus being safer than saflufenacil (80% injury), while exhibiting low toxicity to zebrafish and earthworms. Molecular docking revealed hydrophobic interactions with Leu356, Leu372, and Phe392, along with two unique side-chain hydrogen bonds involving Arg98 and Ser235. Its calculated binding free energy (ΔGcal = −30.579 kcal/mol) was more favorable than that of saflufenacil (−25.386 kcal/mol). These findings identify I18 as a promising, effective, and environmentally compatible PPO inhibitor for weed management. ## DockM8: an all-in-one open-source platform for consensus virtual screening in drug design - Source: Journal of Cheminformatics (journals) - Date: 2026-09-18T00:00:00+00:00 - Categories: Docking & Screening - Authors: Antoine Lacour, Hamza Agha, Anna K. H. Hirsch, Andrea Volkamer - Journal: Journal of Cheminformatics - DOI: 10.1186/s13321-026-01287-2 - Keywords: virtual screening - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Examining Protein Residue-Level Stability Using Theory and Experiment - Source: ACS Omega (journals) - Date: 2026-09-18T00:00:00+00:00 - Authors: Andrew D. Sanders, Rickey Y. Yada, Derek R. Dee - Journal: ACS Omega - DOI: 10.1021/acsomega.6c05138 - Source URL: - Dashboard article: Abstract: Protein behavior, whether evolved in nature or intentionally designed, is governed by the energetics of amino acid interactions that bridge sequence to function. As an extension of the theory of protein folding funnels, local frustration quantifies the optimality of these residue-level interactions relative to alternatives. Although this framework has existed for decades, experimental validation has remained elusive. Past deep mutational scanning datasets enable experimental assessment across nearly 8,000 positions from 178 proteins. Results provide experimental evidence for theoretical predictions regarding the relationship between local frustration and protein sequence, structure, and evolution. An evaluation of three in silico methods shows modest agreement with benchmarks (r = 0.03 to 0.26), with the predominant local frustration predictor (Protein Frustratometer) capturing under 3% of the experimental variance. A recent deep learning model, Pythia, outperformed other tools across all evaluated metrics (r = 0.42). Finally, this scale of empirical data enables an evaluation of the strengths and limitations of energetic measures themselves, inspiring a complementary Sequence Probability INverse (SPIN) framework, which characterizes optimality through a Boltzmann-weighted selection probability within an ensemble of sequences. These findings help provide experimental grounding for the theoretical principles that govern protein sequence energetics. ## Exploring Anti-Breast Cancer Potential and Mechanism of Novel Xanthohumol-Metformin Conjugate by Integrating Network Pharmacology, Molecular Docking and Simulation Approaches - Source: Journal of Computational Biophysics and Chemistry (journals) - Date: 2026-09-18T00:00:00Z - Categories: Docking & Screening, ADMET & Safety - Authors: V. Harish, Sharfuddin Mohd, Goparaju Kavya, B. C. Revansiddappa, Srikanth Jupudi, V. R. Devatharun - Journal: Journal of Computational Biophysics and Chemistry - DOI: 10.1142/s2737416526501255 - External ID: 8a26f4897f892dc0d15f7dfa2238a776e35caa6d - Keywords: Molecular Docking, molecular dynamics, MD simulations, ADME, pharmacokinetic, receptor, density functional theory, DFT - Source URL: - Dashboard article: Abstract: Natural product scaffolds can be combined with repurposed drugs, so called conjugate molecules may exhibit improved pharmacology. We designed a conjugate (XN-MT) of xanthohumol (a prenylated chalcone with anticancer properties) and metformin (an antidiabetic drug with reported anticancer effects) and evaluated its potential against breast cancer (BC) using network-based pharmacology, docking, molecular dynamics (MD) simulation and density functional theory (DFT) Studies. Network pharmacology revealed targets relevant to cell proliferation, apoptosis, CDK4/2/1 signaling, and metabolic regulation and identified pathways were crucial for breast tumorigenesis and chemoresistance. Molecular docking against high-priority targets, like CDK4, PTGS2, and CDK1, predicted favourable binding affinities and plausible interaction modes for the conjugate. MD simulations (100 ns) for the conjugate showed stable ligandprotein interactions and persistent hydrogen bond networks, supporting the conjugates predicted binding stability. ADME, toxicity and DFT analyses provided comprehensive insights into the pharmacokinetic profile and electronic behaviour of xanthohumol, metformin and their conjugate. Moreover, DFT calculations on conjugates showed a lower HOMO-LUMO energy gap, average local ionization, decreased electrostatic potential, lower electron affinity, and higher chemical potential than xanthohumol, metformin, and Palbociclib, indicating greater reactivity and stronger receptor interactions. This integrative in-silico pipeline supports the XN-MT conjugate as a potential candidate for further in-vitro studies and preclinical development in breast cancer therapy targeting CDK4. ## Extracellular signal-regulated kinase 3 forms a nuclear complex with Aly/REF export factor and the splicing factor proline and glutamine rich to exacerbate pathological cardiac remodeling due to pressure overload - Source: Molecular Biomedicine (journals) - Date: 2026-09-18T00:00:00Z - Authors: Wen-Ying Zhou, Feng Wang, Shu-Yan Wang, Li-Guo Wang, Ai-Qun Chen, Hao-Yue Tang, Ya-Peng Chen, Huan Zhang, Xiao-Fei Gao, Juan Zhang, Shao-Liang Chen, Jun-Jie Zhang - Journal: Molecular Biomedicine - DOI: 10.1186/s43556-026-00563-9 - External ID: 45e2b7aff695ea00c85bf32163092e4e1bf29844 - Keywords: virtual screening, kinase, receptor - Source URL: - Dashboard article: Abstract: Pathological cardiac remodeling is a significant contributor to heart failure and mortality. Studies have demonstrated that extracellular signal regulated kinase 3 (ERK3) in cardiac fibroblasts aggravates pressure overload-evoked pathological cardiac remodeling via mitogen activated protein kinase-activated protein kinase-5 (MK5). However, myocardial ERK3 does not activate MK5, thus the role of myocardial ERK3 remains unclear. Through in vivo and in vitro experiments, we revealed that myocardial ERK3 expression increases under pressure overload and the protein accumulates significantly in the nucleus. Using cardiomyocyte-specific Erk3-deficient mice, neonatal rat cardiomyocytes and adult mouse cardiomyocytes, we found that cardiomyocyte-specific Erk3 deficiency ameliorated pressure overload-induced pathological cardiac remodeling in vivo. Co-immunoprecipitation, mass spectrometry, and single nucleus RNA sequencing were used to analyse for the nuclear translocation and downstream signaling mechanisms of ERK3. Specifically, pressure overload enhances the binding of Aly/REF export factor (ALY) to ERK3, resulting in the truncation of the ERK3 C-terminal and its subsequent nuclear translocation to activate the thioredoxin-interacting protein (TXNIP)/NOD-like receptor thermal protein domain associated protein 3 (NLRP3) pathway through the expression of the splicing factor proline and glutamine rich (SFPQ). To screen for targeted drugs, we conducted a virtual screening and identified estrone sulfate as an inhibitor of ALY to counteract hypertrophic effects. Collectively, our findings indicate that estrone sulfate functions as a novel inhibitor of the ALY-ERK3 signaling pathway, potentially serving as a promising therapeutic candidate for the management of pathological cardiac remodeling. ## Extraction of Phenolic Compounds from the Pulp of Myrciaria dubia Using Eco-Friendly Solvents: Extraction Kinetics and Determination of Mass Transfer and Thermodynamic Parameters - Source: ACS Omega (journals) - Date: 2026-09-18T00:00:00+00:00 - Authors: André Gomes Mesquita, Adriny dos Santos Miranda Lobato, Iane Valente Pires, Luiza Helena Meller da Silva, Antonio Manoel da Cruz Rodrigues - Journal: ACS Omega - DOI: 10.1021/acsomega.6c02599 - Keywords: diffusion model - Source URL: - Dashboard article: Abstract: Sustainable extraction of bioactive compounds from Amazonian fruits is essential for green chemistry, but integrated kinetic, thermodynamic, economic, and sustainability assessments for Myrciaria dubia (camu–camu) phenolics remain limited. This study investigated phenolic extraction from camu–camu pulp using natural deep eutectic solvents (NADES) composed of choline chloride and acetic acid, either pure or in binary mixtures with water or ethanol (EtOH), combined with conventional (CE), shear (SE), and ultrasound-assisted extraction (UAE; 420, 490, 560 W). Extraction kinetics, mass transfer parameters, thermodynamics, environmental performance (EcoScale and AGREE metrics), and preliminary economics were evaluated. UAE using NADES + EtOH at 560 W yielded the highest phenolic content (17.11 mg GAE/g). Fick’s diffusion model best described the kinetics (R2 > 0.9523), indicating a diffusion-controlled process, while thermodynamic analysis confirmed an endothermic and spontaneous process. All DES-based systems showed superior environmental performance over CE using 70% (v/v) EtOH, with EcoScale scores of 87.97–92.97 (vs 83.43) and AGREE values of 0.61 (vs 0.52). Preliminary economic assessment indicated extract production costs of US$ 1.83–2.26/kg; NADES-based systems maintained competitive costs while providing higher phenolic content than conventional extraction. Combining NADES + EtOH and UAE represents a promising strategy for enhancing phenolic recovery while maintaining favorable sustainability and economic performance, supporting greener extraction of Amazonian resources. ## HIV-1 drug resistance transmission cluster dynamics among men who have sex with men across Europe: a multi-regional molecular epidemiological study - Source: Scientific Reports (journals) - Date: 2026-09-18T00:00:00+00:00 - Authors: Beatriz Valadas, Cruz S. Sebastião, Mafalda Miranda, Luisa M. Lobo, Inês Alves, Marta Pingarilho, Ana B. Abecasis, Victor Pimentel - Journal: Scientific Reports - DOI: 10.1038/s41598-026-71474-6 - Source URL: - Dashboard article: Abstract: The HIV epidemic among men who have sex with men (MSM) in Europe is sustained by structured transmission networks in which drug resistance mutations (DRMs) spread across geopolitical boundaries. Herein, we applied a systematic molecular epidemiological framework to characterise DRM prevalence, temporal trends, and identify transmission clusters from MSM across European geopolitical regions (Western Europe, Central and Eastern Europe, Southern Europe, and Northern Europe). This was a retrospective molecular epidemiological analysis conducted with 3300 HIV pol sequences from 1983 to 2016. Sequences were aligned with ViralMSA, the maximum likelihood (ML) tree was reconstructed with IQ-TREE, the transmission cluster was inferred with ClusterPicker, and HIVDR was evaluated with Stanford HIVdb. An overall DRM prevalence of 28% (95% CI: 26.5–29.5%) was detected, with NRTIs as the dominant drug class affected (12.8%). A significant decreasing temporal trend was detected (Mann–Kendall τ = − 0.52, p < 0.05), corresponding to an estimated reduction of 1.42 percentage points per year. ML phylogenetic inference followed by transmission cluster detection identified 492 clusters encompassing 49% of sequences; 92% were geographically confined to a single region, while 8% exhibited interregional spread predominantly linking Central and Eastern with Western Europe. Drug-resistant sequences in Western Europe were significantly less likely to belong to a transmission cluster (OR = 0.57; 95% CI: 0.48–0.68; p < 0.001). These findings reveal that resistant variants are actively clustering and spreading across European borders, underscoring the need for coordinated multi-regional molecular surveillance. ## HLA-DRB1 Polymorphisms Modulate the Dynamic Compatibility between the Presentation of EBNA1400–413 and MBP85–99 - Source: ACS Omega (journals) - Date: 2026-09-18T00:00:00+00:00 - Authors: Levy Bueno Alves, Silvana Giuliatti - Journal: ACS Omega - DOI: 10.1021/acsomega.6c06047 - Keywords: molecular dynamics - Source URL: - Dashboard article: Abstract: Multiple sclerosis (MS) shows a strong genetic association with HLA-DRB1 alleles and a consistent epidemiological association with Epstein–Barr virus (EBV) infection. However, it remains unclear how HLA-DRB1 polymorphisms modulate the presentation of autoantigens and viral antigens potentially involved in cross-reactivity. In this study, peptide–HLA (pHLA) complexes formed by six HLA-DRB1 alleles associated with different MS risk profiles were investigated in interaction with CLIP103–117, MBP85–99, and EBNA1400–413. To this end, molecular dynamics simulations were integrated with machine learning models trained using inter-residue distances at the pHLA interface as input attributes, allowing evaluation of how different peptides modulate the conformational stability of the complexes and identification of which interface regions contribute to distinguishing alleles associated with higher and lower risk. The results show that EBNA1400–413 does not globally reproduce the interaction mode of MBP85–99 within the binding groove. Compared with MBP85–99 and CLIP103–117, EBNA1400–413 exhibited fewer intermolecular contacts, greater solvent exposure, and less favorable binding energy. Despite this, in the higher-risk alleles DRB1\*15:01 and DRB1\*15:03, complexes containing EBNA1400–413 showed lower conformational variability and occupied low-energy regions that partially overlapped with those observed for MBP85–99. The machine learning models indicated distinct discriminatory patterns for the autoantigen and the viral antigen. However, in the combined MBP85–99 and EBNA1400–413 model, the most informative pairs converged on the β:70−β:72 region, including β:71, with more pronounced differences in complexes containing EBNA1400–413. This result suggests that β:71 contributes to interface organization in a peptide- and allele-dependent manner. In addition, the proximity between C:R402 of EBNA1400–413 and the region occupied by C:K93 in MBP85–99 suggests a local physicochemical similarity between the complexes, despite the absence of global structural equivalence between the peptides. These findings indicate that EBNA1400–413 does not act as an ideal structural mimic of MBP85–99, but may favor partially compatible conformational states in alleles associated with higher MS risk. Thus, this study proposes a mechanism of functional mimicry dependent on pHLA interface dynamics, in which HLA-DRB1 polymorphisms modulate the presentation of self and viral peptides. This model provides a molecular basis for investigating how the interaction between HLA-associated genetic risk and the immune response to EBV may contribute to autoimmune mechanisms in MS. ## Identification of multi-target lead compounds from iron-stressed Aspergillus species using LC-MS, in silico analysis, and pharmacokinetics profiling. - Source: Natural product research (journals) - Date: 2026-09-18T00:00:00Z - Categories: Docking & Screening - Authors: Dina Mistry, Dhara A. Gamit - Journal: Natural product research - DOI: 10.1080/14786419.2026.2734558 - External ID: a246467f2bdc8388dd64a6bd86bd8f1fe5e7fda7 - Keywords: drug likeness, molecular docking, Pharmacokinetic, kinase - Source URL: - Dashboard article: Abstract: Natural products from soil-derived fungi represent an important source of bioactive compounds with therapeutic potential. In this study, Aspergillus spp. isolated from soil samples were cultured under iron-limited conditions to stimulate siderophore-associated secondary metabolism. Three lead metabolites, Unguisin E, Austalide Q, and Notoamide J, were characterised using LC-MS analysis. Their therapeutic relevance was evaluated through molecular docking against EGFR tyrosine kinase (1M17), PPAR-γ (PDB IDs: 2PRG and 2HNP), and DNA gyrase (1KZN) representing anticancer, antidiabetic and antimicrobial targets. The metabolites exhibited notable binding affinities comparable to standard drugs, including Erlotinib and Rosiglitazone. Austalide Q and Notoamide J showed stable interactions within the active sites of both proteins. Pharmacokinetic and toxicity predictions using SwissADME, pkCSM, and ProTox-3.0 indicated favourable drug-likeness, high intestinal absorption, and no predicted hepatotoxicity or immunotoxicity. Overall, these findings suggest that Aspergillus-derived metabolites may serve as promising candidates for future antimicrobial and therapeutic drug discovery studies. ## Identification of potential SARS-CoV-2 Mpro inhibitors from a Thai natural product database via interaction fingerprint-based virtual screening - Source: Journal of Computer-Aided Molecular Design (journals) - Date: 2026-09-18T00:00:00+00:00 - Categories: Property Prediction - Authors: Aunlika Chimprasit, Guillaume Bret, Supa Hannongbua, Patchreenart Saparpakorn, Didier Rognan - Journal: Journal of Computer-Aided Molecular Design - DOI: 10.1007/s10822-026-00941-z - Keywords: virtual screening - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Integrated In Silico and Experimental Insights into Immunomodulatory Toll-Like Receptor 7/8 Agonists with Anticancer Potential - Source: ACS Pharmacology & Translational Science (journals) - Date: 2026-09-18T00:00:00Z - Categories: Docking & Screening - Authors: Pankaj Kumar, Zahid Bashir Zargar, S. P. Singh, Kushvinder Kumar, Binita Sihag, Shubham Debaje, A. Sangamwar, Pavitra Ranawat, Gurpal Singh, R. P. Barnwal, Deepak B. Salunke, Sandip V. Pawar - Journal: ACS Pharmacology & Translational Science - DOI: 10.1021/acsptsci.6c00263 - External ID: a5d7f7d6402b20e846644236e0ee6c91c1bfcded - Keywords: molecular docking, molecular dynamics, pharmacokinetic, Receptor - Source URL: - Dashboard article: Abstract: Toll-like receptors (TLRs) are key pattern recognition receptors that play a central role in regulating and bridging both innate and adaptive immune responses. Notably, TLR7 and TLR8 agonists have gained significant attention as promising therapeutic candidates because of their ability to activate immune responses and modulate immune signaling pathways for treating viral infections, autoimmune disorders, and cancer. Despite the growing interest in the area of immunotherapy, there remains a substantial gap in the development of new immunomodulators with improved efficacy and safety profiles. The primary objective of the study was to identify and evaluate newly synthesized TLR7/8 agonists using integrated in silico and in vivo strategies. Computational analyses, including molecular docking, molecular dynamics (MD) simulations, pharmacokinetic-pharmacodynamic profiling, and network pharmacology analysis, revealed favorable binding interactions, stability, drug-like properties, and potential targeting of immune- and cancer-related pathways, specifically those relevant to colorectal cancer. In vitro cytotoxicity assays showed that the TLR7/8 agonists exhibit dose-dependent effects in both colorectal cancer cells and RAW 264.7 macrophages. Additionally, the compounds exhibit minimal hemolytic activity, indicating a low potential for systemic toxicity. Furthermore, the cell migration assay demonstrated that several compounds effectively inhibited cancer cell migration, particularly when combined with 5-fluorouracil (5-FU), suggesting potential anticancer and antimetastatic activity. Overall, these findings integrate molecular-level, system-level, and experimental validation approaches to comprehensively evaluate TLR7/8 agonists as potential immunotherapeutic agents. However, further comprehensive in vitro and in vivo investigation studies are necessary to establish the potential of TLR7/8 agonists for use in cancer immunotherapy. ## Integrated AI-Driven Multi-Objective Optimization of Sulfamoylbenzamide-Based HBV Capsid Assembly Modulators: Discovery of 11B-6 with Superior Potency and Drug-like Properties - Source: Journal of Medicinal Chemistry (journals) - Date: 2026-09-18T00:00:00+00:00 - Authors: Shuo Wang, Miaochen Xu, Feiyue Ma, Dazhou Shi, Leda C. Bassit, Mohammad Salman, Harout Ajoyan, Delgerbat Boldbaatar, Tamara McBrayer, Xiaoyu Shi, Shuo Wu, Xinyong Liu, Shujing Xu, Thomas Tu, Raymond F. Schinazi, Yibei Xiao, Peng Zhan - Journal: Journal of Medicinal Chemistry - DOI: 10.1021/acs.jmedchem.6c01701 - Keywords: activity prediction, Multi Objective Optimization, EC50 - Source URL: - Dashboard article: Abstract: Current capsid assembly modulators (CAMs) require improvements in antiviral potency and drug-like properties. Herein, we report a multi-objective optimization workflow for the NVR 3-778 by integrating structure-based fragment growing with systematic medicinal chemistry optimization, activity prediction, and multitask prediction of drug-like properties. This led to 48 novel analogs. Among them, 11B-6 showed superior antiviral activity in HepAD38 cells (EC50 = 0.04 ± 0.02 μM), outperforming NVR 3-778 (EC50 = 0.50 ± 0.17 μM). In HBV-infected HepG2-NTCP cells, 11B-6 inhibited secreted HBV DNA with an EC50 of 3.71 nM, about 134-fold more potent than NVR 3-778 (EC50 = 497.9 nM). Structural biology analyses revealed a distinct, more stable binding mode of 11B-6 at the capsid dimer−dimer interface. 11B-6 also showed lower plasma protein binding than NVR 3-778 (84.3% vs. 100%). These results highlight 11B-6 as a promising next-generation HBV CAM, representing a valuable starting point for advancing its development. ## Integrated multi-omics and experimental analyses reveal VIPR1 as a potential mediator linking environmental bisphenol A exposure to MASH-HCC progression. - Source: Ecotoxicology and environmental safety (journals) - Date: 2026-09-18T00:00:00Z - Categories: Docking & Screening - Authors: Meng-Yue Li, Yan-Ming Yang, Li-Ye Zhong, Li-Yu Zhang, Geng-Luan Liu, Kang-Cong Liang, Zhang Fu, Jun-Jing Zhang, Ning-Ning Li, You-Peng Chen - Journal: Ecotoxicology and environmental safety - DOI: 10.1016/j.ecoenv.2026.120809 - External ID: fd130a1bb38eac99ed679bcdf0ffaf83d7cd204d - Keywords: virtual screening, molecular docking - Source URL: - Dashboard article: Abstract: Metabolic dysfunction-associated steatohepatitis-related liver cancer (MASH-HCC) is an increasingly important subtype of hepatocellular carcinoma, and environmental contaminants have been proposed as potential contributors to its progression. Bisphenol A (BPA), a ubiquitous endocrine-disrupting chemical, has been implicated in metabolic dysfunction and liver injury. However, the molecular mechanisms linking BPA exposure to MASH-HCC remain incompletely understood. In this study, we employed an integrated framework combining transcriptomic analysis, weighted gene co-expression network analysis, machine learning, single-cell RNA sequencing, virtual knockout analysis, DrugReflector-based virtual screening, molecular docking, and experimental validation to investigate BPA-associated molecular alterations during MASH-HCC progression. VIPR1, AURKA, and CYP2C9 were identified as core candidates through integrative multi-omics analysis and machine-learning screening, with SHAP analysis showing the highest contribution of VIPR1 to model prediction. Single-cell and virtual perturbation analyses revealed hepatocyte-enriched expression patterns and prioritized YY1 as a potential upstream regulator of VIPR1. In vitro experiments further demonstrated that BPA exposure reduced nuclear YY1 abundance, accompanied by decreased VIPR1 expression under lipotoxic conditions, whereas YY1 overexpression partially restored VIPR1 expression. Moreover, VIPR1 overexpression attenuated, while VIPR1 knockdown enhanced, malignant-associated cellular phenotypes. Tasisulam was further identified as an exploratory transcriptomic reversal candidate and was associated with increased VIPR1 protein expression. Collectively, these findings identify a VIPR1-centered molecular vulnerability linking BPA exposure with malignant-associated alterations under lipotoxic conditions and provide candidate biomarkers and therapeutic targets for further investigation. ## Interplay of Lattice Distortion and Cation Disorder Governs Li-Ion Transport in Cation-Disordered Rocksalt Cathodes - Source: Journal of the American Chemical Society (journals) - Date: 2026-09-18T00:00:00+00:00 - Categories: ML Potentials - Authors: Zichang Zhang, Lihua Feng, Jiewei Cheng, Peng-Hu Du, Chu-Liang Fu, Xintao Long, Anchun Tang, Longlong Fan, Kang Dong, Haoyu Wu, Weihan Li, Jian Peng, Shuo Wang, Dingguo Xia, Xueliang Sun, Qiang Sun - Journal: Journal of the American Chemical Society - DOI: 10.1021/jacs.6c05378 - Keywords: molecular dynamics - Source URL: - Dashboard article: Abstract: Cation-disordered solids provide a chemically complex landscape in which local environments, lattice responses, and configurational disorder collectively influence ion transport. In cation-disordered rocksalt cathodes, Li+ diffusion has traditionally been interpreted using the static 0-transition-metal (0-TM) percolation rule, which assumes an ideal lattice and often underestimates experimentally accessible capacities. Here, we show that lattice distortion constitutes an essential and previously overlooked chemical degree of freedom that actively reshapes the Li+ percolation networks in disordered oxides. By combining Monte Carlo sampling of cation configurations with molecular dynamics simulations accelerated by machine learning interatomic potentials, we develop a lattice-responsive framework that quantitatively predicts Li+ percolation and electrochemical capacities with deviations from experiment below 5%. Our results reveal a causal coupling between lattice distortion and cation short-range order: enhanced local distortions suppress ordering and activate Li+ migration through nominally inaccessible 1-transition-metal (1-TM) diffusion channels, thereby extending the percolation network beyond the conventional 0-TM paradigm. Guided by this insight, we design and synthesize a multication cation-disordered rocksalt cathode, Li1.2Mn0.2Ti0.2V0.2Mo0.2O2, which exhibits increased lattice distortion, an expanded Li+ percolation network, and a high reversible capacity consistent with theoretical predictions. These findings establish lattice distortion as an active chemical parameter governing ion transport in disordered solids and provide a general design principle for ion-conducting materials. ## Isolation of Bacillus subtilis NJ01 and In Silico Evaluation of Its Antimicrobial Metabolites as Potential Inhibitors of the CRN8 Effector Protein of Phytophthora infestans. - Source: Microbial pathogenesis (journals) - Date: 2026-09-18T00:00:00Z - Categories: ADMET & Safety - Authors: Nafisa Jakia, Alomgir Hossain, Sohana Mehjabin, Shilan S. Saleem, Rashed Zaman, M. Khalekuzzaman, Md Motiur Rahman, M. Matin, Md Asadul Islam - Journal: Microbial pathogenesis - DOI: 10.1016/j.micpath.2026.108836 - External ID: 49b356a13f61e255c54cc09826390a25830bd3af - Keywords: drug likeness, Virtual screening, molecular dynamics, ADMET, molecular features, bioactivity, free energy - Source URL: - Dashboard article: Abstract: Late blight, caused by Phytophthora infestans, is one of the most destructive oomycete diseases affecting global potato and tomato production, leading to annual economic losses exceeding $6 billion. Overreliance on synthetic fungicides has raised serious environmental and resistance-related concerns, highlighting the need for sustainable biological alternatives. This study aimed to isolate and characterize antagonistic rhizobacteria and evaluate their bioactive metabolites as potential inhibitors of the CRN8 effector protein of P. infestans using computational approaches. Six bacterial isolates were recovered from rhizospheric soil collected from the Bangladesh Rice Research Institute (BRRI), Rajshahi, Bangladesh. Among them, Isolate-2 exhibited the strongest antagonistic activity, causing approximately 54% mycelial growth inhibition against P. infestans in dual-culture assays, and was identified as Bacillus subtilis NJ01 based on morphological, biochemical, and 16S rRNA gene sequence analyses. The three-dimensional structure of the CRN8 protein was predicted using AlphaFold2 and validated using PROCHECK, ERRAT2, and Verify3D. Virtual screening of 76 reported B. subtilis-derived antimicrobial compounds identified three compounds: Androst-5-en-3-ol, 4,4-dimethyl-, (3beta)-, 7-O-Malonyl macrolactin A, and Simiarenol, with predicted binding affinities of -8.4, -8.1, and -8.0 kcal/mol, respectively. ADMET (absorption, distribution, metabolism, excretion, and toxicity) profiling indicated favorable drug-likeness and generally favorable predicted toxicity profiles among the selected compounds, although compound-specific differences were observed. POM analysis of five representative compounds (C1-C5) revealed distinct predicted bioactivity profiles across multiple target classes, providing additional insights into their potential molecular features and biological relevance. Furthermore, 100 ns molecular dynamics simulations supported stable protein-ligand interactions, as indicated by RMSD, RMSF, radius of gyration, SASA, principal component analysis (PCA), and free energy landscape analyses. Collectively, these findings suggest that B. subtilis NJ01 possesses promising biocontrol potential and that its reported metabolites warrant further experimental isolation and validation as potential inhibitors of the CRN8 effector protein for the sustainable management of late blight disease. ## LIDR-TB: large language model-integrated platform for traceable drug repurposing in tuberculosis - Source: Journal of Cheminformatics (journals) - Date: 2026-09-18T00:00:00+00:00 - Categories: LLMs & Agents - Authors: ShanShan Hu, Song Jie, Amin Ullah, Jiajia Dong, Xin Zheng, Xingyun Liu, Hui Zong, Jiao Wang, Xiaoyu Li, Bairong Shen - Journal: Journal of Cheminformatics - DOI: 10.1186/s13321-026-01305-3 - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Machine learning-assisted design and FEM simulation of a multi-material metasurface terahertz refractive index sensor - Source: Scientific Reports (journals) - Date: 2026-09-18T00:00:00+00:00 - Authors: Abdulkarem H. M. Almawgani, Nurul Halimatul Asmak Ismail, Samer A. B. Awwad, Pelluce Kabarokolea - Journal: Scientific Reports - DOI: 10.1038/s41598-026-71892-6 - Keywords: regression model - Source URL: - Dashboard article: Abstract: In this paper, the design, numerical simulation, and optimization of label-free terahertz refractive index sensor using periodic metasurface are presented. The sensor is made of an array of 8 unit-cells on a silicon dioxide (SiO₂) substrate. Elliptical resonators with gold (Au), silver (Ag), MXene and phosphorene are placed on top of square coupling layers coated with copper (Cu).The finite element method (FEM) is used in the COMSOL Multiphysics simulation software to calculate the electromagnetic responses. The simulations are based on the Floquet–Bloch periodic boundary conditions and extended Drude–Lorentz dispersion model, which describes the material behavior in the terahertz regime.The four structures are tested for their properties over the range of RIU values of the analyte from 1.3333 to 1.4833. The performance of the detector is judged by sensitivity, figure of merit (FOM), quality factor, detection limit, signal to noise ratio and dynamic range. The fourth performance configuration provides the best performance (1 THz/RIU peak sensitivity, and 26.316 RIU −1 FOM). The performance gradually increases from one configuration to the other, and the achieved results of the evaluated metrics are around 10–15% higher.The machine learning regression model is used to predict the sensor response as a function of angle of incidence (0° to 80°) and square resonator lateral dimension (1.3 to 5.3 μm). For all test cases, the model's R 2 values are greater than 0.998 and RMSE values are less than 0.007, indicating that the response behavior can be quickly predicted without full-wave simulation. The metasurface is intended for label-free bulk refractive-index sensing and tissue dielectric characterization—applications including surgical-margin assessment and ex vivo tissue discrimination in medical diagnostics, together with food-safety and environmental-monitoring tasks that rely on bulk dielectric contrast rather than single-molecule recognition. ## Microplastic exposure is associated with enhanced PI3K-Akt signaling in DSS-induced colitis: integrated bioinformatics analyses and in vivo validation - Source: Frontiers in Pharmacology (journals) - Date: 2026-09-18T00:00:00Z - Categories: Docking & Screening - Authors: Li-Sha Lu, Yuan-Ming Yang, Shun-Yong He, Shao-Gang Huang, Yu-Long Li - Journal: Frontiers in Pharmacology - DOI: 10.3389/fphar.2026.1875549 - External ID: aac135bcad0b5692c47721c52ccdd07a03a56963 - Keywords: molecular docking, receptor - Source URL: - Dashboard article: Abstract: Microplastics (MPs) are widely distributed in the environment and may aggravate ulcerative colitis (UC), but the underlying molecular mechanisms remain unclear. We integrated network toxicology, RNA-seq-based differential expression analysis, molecular docking, GeneMANIA-based functional association analysis, and experimental validation in a murine DSS-induced colitis model. We identified 421 MP-related and 6,452 UC-related targets, including 265 overlapping targets. PIK3CD, PIK3CA, SRC, PTPN11, and EGFR were prioritized as hub targets with stable ligand binding. Enrichment analyses implicated the PI3K-Akt, JAK-STAT, T-cell receptor, and HIF-1 signaling pathways. In vivo experiments showed that MP exposure increased the colonic abundance of these five proteins and enhanced p-PI3K and p-AKT signals in DSS-induced colitis. MP exposure is associated with enhanced PI3K-Akt signaling in experimental colitis. These findings provide preliminary biological support for the PI3K-Akt pathway as a potentially relevant mechanism linking MP exposure to UC aggravation. ## Molecular and biophysical foundations of nano-enabled seed priming and smart nanocarriers for crop stress alleviation - Source: Frontiers in Plant Science (journals) - Date: 2026-09-18T00:00:00Z - Authors: A. Abdulbaki, H. Alsamadany, B. Olayinka, Y. Alzahrani - Journal: Frontiers in Plant Science - DOI: 10.3389/fpls.2026.1954570 - External ID: a8f5d342853e80c38d3e5e768c82321e171eac95 - Keywords: QSAR - Source URL: - Dashboard article: Abstract: Agricultural productivity is increasingly constrained by interacting abiotic stresses, yet translation of nano-enabled seed treatments from controlled experiments to field agriculture remains limited. Seed nanopriming can initiate physiological and molecular priming before radicle emergence, but outcomes depend strongly on nanomaterial identity, particle size and morphology, surface charge, colloidal stability, cargo chemistry, exposure regime, and crop genotype. This review critically examines mesoporous silica nanoparticles, chitosan-based systems, metal–organic frameworks, hybrid/core–shell platforms, and catalytic nanozymes, distinguishing true cargo-delivery carriers from catalytic functional nanomaterials. We integrate interfacial physics with seed-coat and cell-wall barriers, clarifying how zeta potential, surface charge density, hydrodynamic size, porosity, functionalization, and material transformation influence adhesion, uptake, release, and translocation. At the molecular level, we connect nanoparticle–surface interactions with ROS and Ca2+ signaling, membrane-potential changes, MAPK/CDPK cascades, phytohormone cross-talk, transcription-factor activation, antioxidant defenses, and stress-memory processes. We also distinguish transient physiological memory from mitotically persistent and experimentally demonstrated transgenerational inheritance, and critically examine bio–nano combinations in soil, including dissolution, redox transformation, heteroaggregation, and eco-corona formation. Particular emphasis is placed on dose- and material-dependent phytotoxicity, multi-stress cross-tolerance, regulatory status of transgene-free editing, and practical barriers involving reproducibility, cost, scalability, field stability, and environmental safety. Finally, we identify realistic research priorities, including standardized nanoformulation reporting, field-validated dose windows, multi-omics and mechanistic imaging, Nano-QSAR/graph-learning workflows, and safer-by-design criteria for precision seed treatment. ## Multifunctional Defense of Achelura yunnanensis Cocoon: High-Strength Tough Silk Fiber, Biomineral Reinforcement, and Protease Degradation Resistance. - Source: ACS applied materials & interfaces (journals) - Date: 2026-09-18T00:00:00Z - Categories: Docking & Screening, Free Energy & MD - Authors: Neng-Wu Wang, Yu-Ying Wang, Nai-Yong Liu, Wen-Yue Liu, Ya-Wen Wang, Xin-Ying Li, Qi Luo, Xiao-Lu Zhang, Peng-Chao Guo, Yan Zhang, Xin Wang, Qing-You Xia, Zhao-Ming Dong - Journal: ACS applied materials & interfaces - DOI: 10.1021/acsami.6c14733 - External ID: 18e6fcdc35efcea7e10c0815c688eb66262c79ac - Keywords: Molecular docking, enzyme - Source URL: - Dashboard article: Abstract: Natural silks offer combinations of mechanical performance and biological functionality that synthetic materials still struggle to replicate. Here, we report the structural, mechanical, and biochemical characterization of cocoon silk from Achelura yunnanensis (Lepidoptera: Zygaenidae), a moth that builds a leaf-wrapped cocoon with three integrated defense layers. The silk fiber exhibits an average tensile strength of 1038 ± 434 MPa, a toughness of 81 ± 49 MJ m-3, and an elastic modulus of 22.1 ± 8.8 GPa, approximately twice the respective values for Bombyx mori silk. These properties stem from a β-sheet content of 44.0 ± 1.6% and crystallinity of 59.6%, enabled by an anterior silk gland that occupies 55% of the total gland length (vs 11.3% in B. mori) and imposes prolonged shear-driven molecular alignment during spinning. The fibroin heavy chain carries a chimeric motif architecture: silkworm-type (GAGAGSGSGA)n repeats (17.6%) coexist with (A)n segments (31.4%) and (GXGGXGXX)n motifs (14.1%) closely resembling spider dragline silk sequences. On the exposed side of the A. yunnanensis cocoon, abundant calcium oxalate monohydrate crystals (56.1% COM content) boost the specific puncture strength to 28.9 N/mm, more than double the 13.8 N/mm of B. mori. Proteomic profiling identified 36 putative antimicrobial proteins in the cocoon, dominated by trypsin inhibitor-like (TIL) proteins. Fluorescence-based enzyme inhibition assays show that these silk proteins suppress both microbial serine proteases (proteinase K residual activity: 0.7%; subtilisin: 1.8%) and animal digestive proteases (trypsin: 24.5%; chymotrypsin: 57.5%). Molecular docking of the predominant inhibitor TIL1 against all four target proteases yields binding free energies of -9.3 to -15.4 kcal/mol with extensive salt-bridge networks. A. yunnanensis cocoon silk thus functions as a naturally integrated physical-chemical composite: high-strength fibers, selective biomineral reinforcement, and broad-spectrum resistance to protease degradation, providing a model system for bioinspired material design. ## Natural Biodegradable Polymer-Based Microneedles for Controlled Drug Delivery: A Rational Design Framework and Translational Perspectives. - Source: Pharmaceutical development and technology (journals) - Date: 2026-09-18T00:00:00Z - Authors: K. Abuelella, Nermin M. Sheta - Journal: Pharmaceutical development and technology - DOI: 10.1080/10837450.2026.2737199 - External ID: c6596623dd08bcfd20ae77437652f3754e6e3534 - Source URL: - Dashboard article: Abstract: Microneedle (MN) technology has emerged as a promising minimally invasive platform for transdermal drug delivery, enabling efficient transport of therapeutic agents across the stratum corneum while improving patient compliance. In recent years, natural biodegradable polymers have gained increasing attention in microneedle fabrication due to their biocompatibility, biodegradability, and functional versatility. This review provides a comprehensive overview of recent advances in natural polymer-based microneedle systems, including material selection, fabrication strategies, and mechanisms of controlled drug release. Importantly, beyond conventional descriptive analysis, this work introduces a rational design framework that systematically integrates therapeutic objectives, polymer properties, mechanical performance, and release kinetics to guide the development of optimized microneedle platforms. The review further addresses key translational challenges, including mechanical limitations, manufacturing scalability, stability concerns, and regulatory considerations. In addition, emerging trends such as stimuli-responsive systems, nanocarrier integration, and personalized drug delivery approaches are discussed. Notably, data-driven strategies, including artificial intelligence (AI) and machine learning, are highlighted as promising tools for predicting system performance, optimizing formulation parameters, and accelerating the development of next-generation microneedle-based drug delivery systems. Overall, this work shifts the paradigm from empirical formulation toward a predictive, engineering-driven design strategy, offering valuable insights to advance the clinical translation of smart microneedle technologies. ## Neurodevelopmental Toxicity of Tris(2,4-di- tert -Butylphenyl)phosphate: Mechanistic Insights from Laboratory Models and Implications for Human Health - Source: Environmental Science & Technology (journals) - Date: 2026-09-18T00:00:00Z - Categories: Docking & Screening - Authors: Jian Li, Hongliang Ji, Yun Liu, Zi-Yue Li, Yuan Fang, Na Li, Kaifeng Rao, Ling-Yu Zhang, Yun-Jiang Yu - Journal: Environmental Science & Technology - DOI: 10.1021/acs.est.6c07562 - External ID: a4f2b0e3bd2661e8b891f1225532f3f05bf4c2b5 - Keywords: Molecular docking, binding affinity - Source URL: - Dashboard article: Abstract: Organophosphite antioxidants (OPAs) and their transformation products, novel organophosphate esters (NOPEs), are ubiquitous environmental contaminants, yet their toxicological profiles remain poorly understood. This study systematically characterized the neurodevelopmental toxicity of these compounds and evaluated their associated environmental risks within the adverse outcome pathway framework. Molecular docking and radioligand binding assays identified integrin αvβ3 as a primary molecular target for all tested OPAs and NOPEs, with tris(2,4-di-tert-butylphenyl)phosphate (AO168═O) exhibiting high binding affinity. Exposure to AO168═O at environmentally relevant concentrations (≥50 μg/L) triggered significant behavioral deficits in zebrafish larvae, including reduced locomotion, altered exploratory preference, and irregular locomotor patterning. In vivo imaging and molecular analyses revealed that AO168═O induced motor neuron abnormalities and suppressed neurogenesis markers. Transcriptomic profiling elucidated that AO168═O disrupted MAPK and calcium signaling pathways via integrin αvβ3 targeting, driving the observed neurotoxicity. Crucially, the derived integrin αvβ3 binding threshold (0.7 μg/L) is comparable to reported environmental concentrations, providing a critical early warning for aquatic ecosystems. Furthermore, “Genes-to-Pathways” analysis demonstrated high conservation of this toxicological pathway between zebrafish and humans, highlighting a potential human health risk. These findings deliver strong evidence for the hazard assessment of AO168═O and offer valuable insights for managing other emerging NOPEs. ## On the Minimization of Graph Counterfactual Explanations: Theory and a Local Bounded Search Algorithm - Source: Machine Learning (journals) - Date: 2026-09-18T00:00:00Z - Authors: Rodrigo García, Mario Alfonso Prado-Romero, Francesco Gullo, G. Stilo - Journal: Machine Learning - DOI: 10.1007/s10994-026-07157-0 - External ID: 5315e958fb3bd8f850d0da5068b17150050c1496 - Source URL: - Dashboard article: Abstract: Graph counterfactual explainability (GCE) addresses the interpretability limitations of opaque machine learning models on graph-structured data by producing graph counterfactuals (GCs): alternative graphs that remain maximally similar to a given instance while inducing a different model prediction. State-of-the-art GCE methods usually follow a generate-and-minimize pipeline: first, generate a valid counterfactual, then refine it to be closer to the original graph. While generation is well-studied, minimization still lacks a formal definition, complexity analysis, and general-purpose algorithms; existing solutions are either simple random edge-swap heuristics or tightly coupled to specific generators, limiting effectiveness and generality. In this work, we address a key gap in GCE by providing the first principled study of minimizing a given valid graph counterfactual. We formalize the task as an optimization problem and prove it is NP-hard. We then propose a decoupled generate-and-minimize framework and introduce Local Bounded Search (LBS), a model-agnostic heuristic that refines any valid counterfactual via constrained structural and attribute edits to reduce dissimilarity while preserving validity. Across nine synthetic and real-world datasets (molecular, biomedical, social), LBS reduces structural edit distance by up to 98% from the initial counterfactual and consistently outperforms existing refinement heuristics. ## Pinocembrin alleviates tubular inflammatory fibrosis and glomerular mesangial lesions in diabetic nephropathy through HSP90AB1/TGF-β1/Smad3 and VEGFR/ERK signaling - Source: Applied Biological Chemistry (journals) - Date: 2026-09-18T00:00:00Z - Categories: Docking & Screening - Authors: Yu-Jie Wang, Jia-Jie Wang, Mao-Xuan Cai, Jie Ren, Fang Jia, Wen-Jie Jiang, Dan Su, Hao Yang, Gui-Lin Chen - Journal: Applied Biological Chemistry - DOI: 10.1186/s13765-026-01136-8 - External ID: a373406e28e9bba849dbd8d9c00ed0658a8cc568 - Keywords: Molecular docking - Source URL: - Dashboard article: Abstract: Diabetic nephropathy (DN) ranks as the primary contributor to end-stage renal disease across the globe, characterized by progressive glomerular mesangial proliferation and tubular interstitial fibrosis, which poses a severe threat to public health. Pinocembrin is a major flavonoid compound that has been reported to exhibit a variety of prominent biological effects. However, its dual regulatory effects on glomerular and tubular injuries in DN remain unclear. The present study aims to investigate the potential role of pinocembrin in alleviating the progression of DN, providing a combination therapeutic strategy. In vivo experiments were performed to verify its efficacy in improving renal injury and fibrosis of pinocembrin in Lepdb/Lepdb (db/db) mice. Pinocembrin-related targets were retrieved from Traditional Chinese Medicine Systems Pharmacology (TCMSP), Comparative Toxigenomics Database (CTD), Swiss Target Prediction, and Symmap databases. Differentially expressed genes in DN tubules and glomeruli were obtained from Gene Expression Omnibus (GEO) dataset GSE30122. Overlapping targets between pinocembrin and DN tissues were identified, followed by Protein-Protein Interaction (PPI) network construction, Gene Ontology (GO) annotation, and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis. Molecular docking predicted potential binding interactions between pinocembrin and key targets. In vitro experiments demonstrated that in HG-induced HK-2 cells, pinocembrin might target HSP90AB1 to suppress inflammatory response and epithelial-mesenchymal transition, thereby alleviating tubular fibrosis injury. In HG-treated SV40 cells, pinocembrin downregulated expression of VEGFR, thereby inhibiting cell proliferation and extracellular matrix deposition to improve glomerular pathological damage. This study clarifies the dual protective mechanism of pinocembrin on DN via regulatory pathways. These findings provide experimental foundation and theoretical insight for the application of pinocembrin as a functional food ingredient for the prevention and treatment of DN. ## Pose Reproduction, Cross Docking, Database Enrichment, and Reverse Docking Benchmarks for DOCK6, Vina, and AutoDock 4 with Emphasis on Approved Drugs - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-18T00:00:00+00:00 - Categories: Docking & Screening - Authors: Carissa P. Corbo, Scott R. Laverty, Robert C. Rizzo - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c01375 - Keywords: AutoDock, AutoDock Vina - Source URL: - Dashboard article: Abstract: High-quality test sets to evaluate docking software help the community understand strengths and weaknesses of different methods and programs. In this work, we introduce SB2025, a calculation-ready set of 1,244 protein-ligand complexes derived from the PDB, curated using literature-based ligand protonation states, and freely available in DOCK6, AutoDock Vina, and AutoDock 4 (AD4) formats. SB2025 includes 322 unique proteins, 17 major protein classes, and ligands with a wide range of properties and flexibilities, making it a broad and challenging test set. FARMA2025, a specific subset of SB2025, contains 165 approved drugs bound to their pharmacological targets and closely mirrors the distribution of FDA drug target classes for small molecules through 2023 (r2 = 0.72). Comprehensive benchmarking was subsequently performed to evaluate pose reproduction (PR), cross-docking (CD), database enrichment (DE), and reverse docking (RD) under standardized conditions. Averaged PR success rate across triplicate runs for DOCK6 was slightly improved compared to Vina (followed by AD4) using both the large SB2025 test set and the more drug-focused FARMA2025 test set. Conversely, CD matrix success rates for AD4 were more favorable than DOCK6 or Vina across 11 protein families. The programs all yielded substantial early DE; however, the number of active ligands in common between the programs within the top 1% of the database, with one exception, was surprisingly low. Challenging RD tests using FARMA2025, which gauge the ability to correctly rank the cognate structure best, indicate that generating the correct ligand pose is necessary but not sufficient. SB2025, FARMA2025, and example scripts and protocols for benchmarking are available at ringo.ams.stonybrook.edu (downloads) and github.com/rizzolab. ## Potential links between gut microbiota-derived propionate and succinate and CXCL8, IL1B, and IL6 in NMOSD: An integrated network pharmacology and mendelian randomization study. - Source: Journal of neuroimmunology (journals) - Date: 2026-09-18T00:00:00Z - Categories: Docking & Screening - Authors: Qing Lin, Yan-Mei Liu, R. He, Zu-Biao Song, Jia-Hui Liang, Wei-Xi Zhang - Journal: Journal of neuroimmunology - DOI: 10.1016/j.jneuroim.2026.579101 - External ID: 2489442d072aae97415bcd248fb276293a4ea569 - Keywords: drug likeness, molecular docking, receptor - Source URL: - Dashboard article: Abstract: BACKGROUND Neuromyelitis optica spectrum disorder (NMOSD) is a severe autoimmune demyelinating disease of the central nervous system. Gut microbiota metabolites may participate in NMOSD pathogenesis, but their biological relevance and underlying molecular links remain unclear. METHODS Candidate key targets of gut microbiota metabolites in NMOSD were obtained from the intersection of public database targets (human gut, NMOSD, and metabolite targets). Functional enrichment and regulatory network analyses were conducted. Microbiota-metabolite-target-signaling pathway (MMTS) network, Mendelian randomization (MR) analysis, and metabolite evaluations (drug-likeness, toxicology, and molecular docking) were performed. RESULTS CXCL8, IL1B, and IL6 were identified as candidate key targets. Notably, they were enriched in pathways related to immune regulation and inflammatory responses, such as "NOD-like receptor signaling pathway", "Toll-like receptor signaling pathway", and "IL-17 signaling pathway". Predicted regulatory factors included hsa-miR-191-5p and hsa-let-7c-5p. Additionally, Barnesiella intestinihominis, Ruminococcus bromii, and Ruminococcus champanellensis were identified as key microbiota. Their associated key metabolites, including propionate and succinate, exhibited favorable drug-likeness, low toxicity, and promising binding affinities to the candidate key targets. CONCLUSION This study suggests that gut microbiota metabolites may affect NMOSD through CXCL8, IL1B, IL6, and immune-inflammatory pathways. Propionate and succinate were identified as candidate metabolites requiring experimental validation of their effects on these targets. ## Public Database Leakage Distorts Model Rankings in Real-Spectra NMR Structure Elucidation - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-18T00:00:00+00:00 - Categories: Property Prediction, Spectra & Analytical - Authors: Zihan Zhang, Mengxi Chen, Xuezhou Zhao, Yutao Guo, Dan Wu - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c02552 - Source URL: - Dashboard article: Abstract: Sequence models that translate NMR spectra into molecular structures report up to 96% top-1 accuracy, but they are pretrained on simulated spectra and tested on public databases that can overlap those corpora. We build a leakage-controlled benchmark on nmrshiftdb2 and find that 26.9% of public molecules are exact training-corpus matches under a stated identity rule, with substantial training-set proximity remaining after exact-match removal. On the wider exact-match-removed real-spectrum cohort, released-model top-1 accuracy is 24.4% (95% CI 23.3–25.6%). A training-free rerank using molecular formula and 13C-peak count consistency raises it to 31.1% (29.9–32.4%), supporting leakage-controlled evaluation and decode-time consistency checks before claims of experimental generalization for simulation-pretrained inverse models. ## Rational Coformer Screening of Multicomponent Crystals of Abiraterone Acetate Using Machine Learning (ML) and Multicomponent Hydrogen Bond Propensity (MCHBP): A Comprehensive Structural and Computational Investigation - Source: Crystal Growth & Design (journals) - Date: 2026-09-18T00:00:00Z - Categories: Cheminformatics - Authors: K. Potla, Abhishek Sharma, Hemanth Kongara, Saurabh Srivastava, Amol G. Dikundwar - Journal: Crystal Growth & Design - DOI: 10.1021/acs.cgd.6c00806 - External ID: 0b6e01da9bf4fb78097f65faaccec723f21ea8f4 - Keywords: RDKit, gradient boosting, force field, pharmacokinetic, molecular descriptors, density functional theory, DFT, B3LYP - Source URL: - Dashboard article: Abstract: Multicomponent crystalline systems offer a viable approach to improving the physicochemical and pharmacokinetic properties of drugs, thereby enabling the rational optimization of solid forms through coformer identification. The conventional process relies on empirical knowledge and chemical intuition. To overcome traditional challenges, we developed a novel predictive framework that combines machine learning (ML) with molecular complementarity hydrogen-bond propensity (MCHBP) for efficient screening of coformers. The effectiveness of the method was confirmed in a proof-of-concept study of the pharmaceutical multicomponent system for abiraterone acetate (AA). The success of this process lies in the generation of high-quality three-dimensional molecular descriptors derived from accurate structures. The novelty of this study is that it uniquely obtains high-quality 3D descriptors using optimized structures with the MMFF94 force field. For each molecule, 1,826 three-dimensional Mordred descriptors were generated with RDKit software. We used two types of descriptor fusion strategies: concatenation (M1) and summation (M2). The performance of the ten machine learning models was evaluated; gradient boosting performed well among the other nine tested models. The predicted API-coformer pairs were ranked in descending order of their probabilities. However, the predicted probabilities of the top 20 candidates were very close, making further prioritization difficult. To refine the selection, we used the MCHBP ranking, which enhanced the matching of hydrogen-bond networks. Following the MCHBP analysis, we selected the top eight unreported coformers for experimental investigation. From these, seven new multicomponent phases were identified, and six multicomponent crystals of AA were confirmed through SCXRD: AA−2HBA, AA−23DHBA, and AA−24DHBA as cocrystals and AA−25DHBA, AA−26DHBA, and AA−246THBA·H2O as salts. The type and nature of intermolecular interactions and their contributions were analyzed by Hirshfeld surface and fingerprint analyses. The electronic structure, charge transfer, and hydrogen bonding were analyzed by density functional theory at the DFT/B3LYP/6-311++G(d,p) level. The cocrystal and salt forms were differentiated using the quantum theory of atoms in molecules (QTAIM) and natural bonding orbital (NBO) analysis. Noncovalent interactions were analyzed by noncovalent interaction (NCI) and reduced density gradient (RDG) graphical user interfaces. ## Resolving Positional Markush Structures for Chemical Reaction Extraction - Source: ChemRxiv (preprints) - Date: 2026-09-18T00:00:00Z - Categories: Cheminformatics - Authors: Ulrick Fineddie Randriharimanamizara, Atif Anwer, Julien Roger, Fabrice Meriaudeau, Paul Fleurat-Lessard - DOI: 10.26434/chemrxiv.15009037/v1 - External ID: 10.26434/chemrxiv.15009037/v1 - Keywords: Markush Structures, Markush, molecular representations - Source URL: - Dashboard article: Abstract: The automated extraction of chemical reaction data from scientific literature remains very challenging mainly due to the visual complexity and heterogeneity of published graphical schemes. In particular, substrate scopes containing positional Markush structures pose a significant challenge for current Optical Chemical Structure Recognition (OCSR) and reaction-extraction systems, as their resolution requires combining molecular recognition with precise spatial reasoning. Herein, we introduce a deterministic, coordinate-based extraction pipeline designed to resolve positional Markush structures in reaction schemes. Building upon a fine-tuned version of the graph-prediction OCSR module MolScribe, our approach uses deterministic geometric rules based on 2D molecular coordinates to infer R-group attachment points and generate topologically valid molecular representations. To evaluate this approach, we introduce two curated benchmarks: MIPub-2k (Molecular Images from Publication - Université Bourgogne Europe) comprising 2,128 literature-derived molecular image crops for OCSR evaluation and MIReactPub (Markush template-based Image of Reaction from Publication – Université Bourgogne Europe) comprising 450 reactions from 23 schemes containing positional Markush structures for reaction-level assessment. The complete pipeline achieves an end-to-end F1 score of 0.46, increasing to 0.54 for schemes without textual-dependent ambiguities. Slightly reorganizing figures doubles the F1 score, demonstrating the robustness of the underlying geometric resolution while highlighting the substantial impact of visual heterogeneity in published literature on end-to-end performance. By explicitly encoding chemical and spatial constraints, our proposed approach provides a transparent and reproducible framework for resolving complex molecular representations from literature images. The resulting open-source pipeline and benchmarks provide a specialized component that can be further integrated into automated workflows for large-scale chemical information extraction. ## Scale Management Technologies for Production Enhancement in Oilfields: Mechanisms, Inhibitor Chemistry, Modeling, and Future Perspectives - Source: ChemEngineering (journals) - Date: 2026-09-18T00:00:00Z - Authors: S. Ahmadi, A. Khormali - Journal: ChemEngineering - DOI: 10.3390/chemengineering10090112 - External ID: 164d15e9ab043549b6542c3df43c16e204fb1e33 - Source URL: - Dashboard article: Abstract: Mineral scale deposition is one of the most persistent flow assurance challenges in the oil and gas industry, causing formation damage, reduced injectivity and productivity, equipment fouling, pipeline blockage, and substantial economic losses. The increasing application of seawater injection, produced-water reinjection, and enhanced oil recovery (EOR) techniques has intensified scaling problems by promoting the mixing of incompatible waters and altering reservoir geochemistry. Consequently, the development of efficient scale management strategies has become essential for maintaining production performance and ensuring the long-term integrity of oilfield assets. This review comprehensively examines the mechanisms of scale formation, the physicochemical and operational factors governing mineral precipitation, and recent advances in scale inhibition technologies for production enhancement. The review discusses the characteristics and formation mechanisms of the major oilfield scales, including carbonate, sulfate, silica, iron-containing, and mixed mineral deposits, together with their effects on reservoir permeability and production facilities. Conventional phosphonate- and polymer-based inhibitors are critically evaluated alongside emerging environmentally friendly inhibitors, nanotechnology-assisted formulations, and controlled-release squeeze treatment systems. Furthermore, laboratory evaluation techniques, adsorption and coreflooding studies, thermodynamic and kinetic modeling, molecular simulations, and artificial intelligence-based predictive methods are reviewed to demonstrate their roles in improving inhibitor design, scale prediction, and treatment optimization. Recent developments in machine learning, digital twins, and intelligent optimization algorithms are also highlighted as enabling technologies for next-generation scale management. Finally, current research challenges and future perspectives are discussed, emphasizing sustainable inhibitor development, integrated experimental and computational approaches, and real-time predictive monitoring systems. By integrating advances in chemistry, materials science, computational modeling, and petroleum engineering, this review provides a comprehensive framework for understanding and implementing effective scale management strategies to enhance hydrocarbon production, reduce operational costs, and improve the sustainability of oilfield operations. ## Several multiple sequence alignment-perturbing methods enhance AlphaFold3 sampling of alternative protein states - Source: Communications Chemistry (journals) - Date: 2026-09-18T00:00:00+00:00 - Authors: Samuel Eriksson Lidbrink, Ivan Nissen, Rebecca J. Howard, Jonathan Kenichi Ahrlind, Erik Lindahl - Journal: Communications Chemistry - DOI: 10.1038/s42004-026-02198-x - Keywords: AlphaFold3, AF3 - Source URL: - Dashboard article: Abstract: Protein function often involves multiple conformational states. Several multiple sequence alignment-perturbing strategies, including stochastic subsampling, clustering, and column masking, have been shown to enhance AlphaFold2 (AF2) sampling of alternative protein states. Here, we evaluate these strategies on AlphaFold3 (AF3) and compare their performance with the BioEmu Boltzmann sampling model on 107 proteins with multiple experimentally solved conformational states. We find that unperturbed AF3 samples alternative states with significantly higher TM-scores compared to AF2 and comparable to BioEmu. In particular, all MSA perturbation methods improve AF3 sampling at a statistically significant level, improving the top 1% TM-score by at least 0.05 in approximately 20% of cases each, while rarely worsening the performance. Furthermore, we find that different choices of amino acid masks can improve column-masked AF3 sampling for specific targets. Our results highlight how MSA perturbations remain relevant in AF3, providing a useful tool for understanding dynamic biological processes. ## Single-cell transcriptomics combined with Mendelian randomization reveals the pivotal role of WDR83OS in clear cell renal cell carcinoma - Source: Medicine (journals) - Date: 2026-09-18T00:00:00Z - Categories: Docking & Screening - Authors: Le-Ting Xiao, Aimureguli Yusan, Yong-De Cao, Fang Yue, Jia-Xin Feng, Hua-Yan Wu, Ling-Li Sun, Jun-Bo Qiu, Yin-Hao Xiao, Shi-Lin Zhang - Journal: Medicine - DOI: 10.1097/MD.0000000000050539 - External ID: 791fae4f17664b6b1cdc86cd5b6bb651cbcc66f9 - Keywords: Molecular docking, binding affinity - Source URL: - Dashboard article: Abstract: This study aimed to comprehensively investigate the molecular mechanisms of clear cell renal cell carcinoma (ccRCC), identify key cellular subpopulations and genes, develop an effective diagnostic model, and screen potential targeted therapies for ccRCC. We analyzed single-cell transcriptomic sequencing data to identify the major cellular subpopulations in ccRCC. High-dimensional weighted gene co-expression network analysis and multiple machine learning algorithms were used to identify key genes and develop a diagnostic model. Two-sample Mendelian randomization analysis was performed to assess causality. Molecular docking was used to identify a candidate therapeutic agent. Data processing was conducted using R and Python. The proportion of endothelial cells was significantly higher in ccRCC (P < .001). High-dimensional weighted gene co-expression network analysis showed that the pink module was closely associated with endothelial cells. Univariate logistic regression and Least Absolute Shrinkage and Selection Operator identified 11 key genes: WDR83OS, TMA7, PFDN5, DSTN, PHPT1, HMGN3, TMSB10, RPL27A, RPL23A, RPL15, and RPL27. Based on these genes, a diagnostic model for ccRCC was developed using multiple machine learning algorithms and achieved an area under the receiver operating characteristic curve of 0.960. In addition, 2-sample Mendelian randomization analysis supported a causal association between WDR83OS and ccRCC (inverse-variance weighted: odds ratio = 1.160, P = .033). Molecular docking indicated that oxyphenbutazone had a high binding affinity for WDR83OS, with a binding energy of −7.124 kcal/mol. By integrating multiple bioinformatic approaches, this study identified key cellular populations and genes in ccRCC, developed a reliable diagnostic model, and highlighted WDR83OS as a potentially important therapeutic target. ## Synthesis of Novel Benzoxazole–Piperazine Hybrids as Anticancer Leads: Biological, Molecular Docking, and ADMET Evaluations - Source: Chemistry Africa (journals) - Date: 2026-09-18T00:00:00Z - Categories: Docking & Screening, ADMET & Safety - Authors: Mahmoud A. Al-Sha'er, A. H. Abdullah, Almeqdad Y. Habashneh, Balakumar Chandrasekaran - Journal: Chemistry Africa - DOI: 10.1007/s42250-026-01846-y - External ID: d4772e1020944d3056a6ac9bcbae6169923b3699 - Keywords: Molecular Docking, ADMET - Source URL: - Dashboard article: - Abstract: not stored for this record. ## The bitterness behind sweetness: an integrated 3D-QSAR and molecular modeling approach elucidates the structure determinants of taste purity in natural chalcones. - Source: Food & function (journals) - Date: 2026-09-18T00:00:00Z - Categories: Cheminformatics, Property Prediction, Docking & Screening - Authors: Shu-Ting Lai, Lu-Fang Chen, Qi-Fei Long, Ling-Xuan Zhang, Yang Hu, Jia Liu, Hui Ni, Feng Chen, Fan He - Journal: Food & function - DOI: 10.1039/d6fo02582j - External ID: 27ec017881123c7e7f230c7e6111128c94b275d1 - Keywords: QSAR, Molecular docking, Molecular Similarity Index - Source URL: - Dashboard article: Abstract: The bitter and astringent properties of plant-derived chalcone compounds result in lower sweet taste purity compared to that of sugars such as sucrose, thereby limiting their applications. To elucidate the relationship between the structure of chalcones and their sweet taste purity, a 3D-QSAR model was established according to the sensory experiments that evaluated the sweetness and aftertaste of 22 chalcone compounds. Based on Comparative Molecular Field Analysis (CoMFA) and Comparative Molecular Similarity Index Analysis (CoMSIA), key sites and functional groups affecting the sweet taste purity of chalcones were identified. The model indicated that the sweet taste purity of chalcone compounds can be enhanced by introducing negative charges and hydrophilic groups at the C2 position of ring A, hydrophilic groups at positions C3 and C4 of ring A, and small-volume groups at position C6 of ring A, as well as introducing hydrophilic groups on ring B and large-volume groups at positions C4' and C5' of ring B. Furthermore, the chalcones compatible with the model were selected for sensory evaluations of the sweet taste purity. The results were consistent with model predictions, thereby validating the efficacy of the 3D-QSAR model. Molecular docking studies further corroborated the model's reliability. This study establishes a 3D-QSAR model for evaluating sweet taste purity scores, thereby enhancing the sensory profile of chalcone compounds. It provides novel design concepts for developing sweeteners with high sweet taste purity and offers the food industry access to high-performance sweetener options. ## Molecular Geometry Understanding Has Unintendedly Emerged in Frontier Large Language Models - Source: arXiv (preprints) - Date: 2026-09-17T16:42:41Z - Categories: LLMs & Agents - Authors: Gregorii A. Semakin, Timofey V. Losev, Ilya V. Prolomov, Stepan N. Ostarkov, Igor V. Alabugin, Michael G. Medvedev - External ID: 2609.20666v1 - Keywords: LLMs, LLM, Claude, GPT, DFT, Force Field, force fields - Source URL: - Dashboard article: - PDF: - Code: Abstract: Large language models (LLMs) have already shown strong capabilities in solving complex chemical problems expressed in natural language. Yet, many tasks performed by chemists require understanding of 3D structures of chemical compounds and reasoning about them -- abilities, which, to the best of our knowledge, were neither intended by LLM developers nor tested in contemporary models. Such understanding is essential for autonomous molecular discovery and drug development, which is one of the Holy Grails of AI application to chemistry. We tested this capability in modern LLMs on 810 geometries and DFT energies of 27 small organic molecules. Strikingly, while models released before July 2025 struggled to rank conformers by stability, many recent frontier models, including GPT-5.6 Sol, Kimi K3, and Gemini 3.6 Flash achieved competitive accuracy, outperforming the Universal Force Field, and GPT-6 Astra and Claude Opus 5 came very close to a modern GFN-FF force field. Analysis of models' explanations for their rankings suggests that they perform best for molecules with well-defined intramolecular interactions -- hydrogen bonds -- which are well quantified in scientific language; at the same time all models except the two newest -- GPT-6 Astra and Claude Opus 5 -- struggle with molecules governed by loosely defined concepts (e.g., ring strain), where force fields excel, suggesting that the scientific language itself might impose constraints on LLMs. Notably, a model's ability to rank conformers is strongly associated with its performance on scientific, coding, and abstract-reasoning benchmarks, suggesting that it emerged unintendedly from models training on linked but conceptually different tasks. Thanks to this generalization of mere scientific knowledge into understanding molecular structures, current frontier models offer a realistic starting point for AI-driven molecular design. ## Truncated automatic sparse differentiation for machine learning interatomic potentials - Source: arXiv (preprints) - Date: 2026-09-17T14:53:14Z - Categories: ML Potentials - Authors: Marcel F. Langer, Adrian Hill, Michele Ceriotti - External ID: 2609.20510v1 - Keywords: MLIP, message passing, molecular dynamics - Source URL: - Dashboard article: - PDF: - Code: Abstract: Machine learning interatomic potentials (MLIPs) learn the mapping from atomic positions to potential energy. The forces, the negative gradient of this energy, drive molecular dynamics and are readily obtained using automatic differentiation. Higher-order derivatives, most notably the Hessian, describe collective motion and allow the direct prediction of experimental observables, but are considered computationally inaccessible for large systems. We suggest a solution: in physical systems, interactions decay with distance, and most MLIPs build on this locality through message passing up to a finite receptive field. This implies both sparsity of higher-order derivatives and their decay with distance. This structure can be exploited using automatic sparse differentiation (ASD). We explain how to compute the sparsity pattern for MLIP derivatives and demonstrate that, for multiple foundation MLIPs, ASD computes full Hessians of large porous materials exactly, but with modest speedups at best. The larger gains come from truncated ASD: discarding small, but nonzero, Hessian entries between distant atoms yields order-of-magnitude speedups with negligible impact on predicted observables. ## Correlation-Free Transition Path Sampling through Shooting Point Generation Guided by Committor Learning - Source: arXiv (preprints) - Date: 2026-09-17T14:24:14Z - Categories: Targets & Structures, Free Energy & MD - Authors: Maximilian Negedly, Sebastian Falkner, Alessandro Coretti, Christoph Dellago - External ID: 2609.20461v1 - Source URL: - Dashboard article: - PDF: - Code: Abstract: Studying the dynamical behavior of a system often depends on characterizing how it transitions between long-lived states. Because such transitions are rare, observing them usually requires specialized enhanced sampling techniques. Transition Path Sampling (TPS) is a well-established method for generating reactive trajectories, which is simple to implement and does not require the definition of a preconceived reaction coordinate. However, its efficiency is limited by its sequential nature and the resulting correlations between sampled paths. Previous work addressed this limitation by combining TPS with a sampling scheme based on conditioned Boltzmann Generators, a generative machine learning model capable of sampling a given target probability distribution. This approach produces uncorrelated transition paths but relies on an accurate reaction coordinate, which is rarely known in advance. Building on recent advances in committor learning, specifically on the Artificial Intelligence for Molecular Mechanism Discovery (AIMMD) method, in this work we introduce GenAIMMD, an iterative algorithm that actively and self-consistently learns the ideal reaction coordinate (the committor) and trains a conditioned Boltzmann Generator to sample from arbitrary bias windows along it. GenAIMMD thereby provides a correlation-free and fully parallelizable path sampling scheme that does not require prior knowledge of the system's transition mechanism. We apply GenAIMMD to a two-dimensional toy model and a higher-dimensional polymer system. In both cases, GenAIMMD succeeds in training the Boltzmann Generator and learning the committor. Benchmark results show a substantial increase in performance compared to standard TPS. ## Pretrained Medical Representations for the Practical Screening of Drug Repositioning Candidates - Source: arXiv (preprints) - Date: 2026-09-17T08:19:34Z - Categories: Property Prediction - Authors: Yuhei Fujioka, Daitaro Misawa, Shingo Fukuma - External ID: 2609.19865v1 - Keywords: BERT - Source URL: - Dashboard article: - PDF: Abstract: Representation learning from medical code sequences in electronic health records and medical claims data has been successful in various clinical applications, such as those regarding disease prediction. However, significant challenges remain in extending this approach to the discovery of scientific hypotheses. One reason is that many existing BERT-based models fail to adequately capture the hierarchical structure of medical codes and the complex interactions between diagnoses and treatments. To address these limitations, we propose a new unified pre-training framework that explicitly integrates hierarchical sub-token aggregation, partial masking, and cross-reference mechanisms. The proposed model consistently outperformed existing methods on both pre-training objectives and downstream clinical event prediction tasks, including the onset of dementia and hospitalization. We also conducted an in silico drug repositioning case study targeting Alzheimer's disease. In the hypothesis generation step, our approach successfully rediscovered known promising drugs in a data-driven manner without relying on such external knowledge sources as the literature. Subsequently, in the hypothesis prioritization step, we introduced a Task-Adaptive Representation Approach to alleviate the over-encoding of historical prescription information within diagnostic vectors, enabling the robust prioritization of generated hypotheses. This study establishes an exploratory screening workflow for hypothesis generation and prioritization based on observational associations. Importantly, this framework is not intended to provide causal evidence, but rather to identify promising candidates for subsequent rigorous causal inference. Overall, this study demonstrates that domain-informed representation learning combined with task-adaptive representation control can enable a practical hypothesis discovery workflow. ## Predicting molecular structure from sparse NMR data, proof-of-concept using a graph neural network - Source: Digital Discovery (journals) - Date: 2026-09-17T08:03:22Z - Authors: Ben Honoré, Calvin Yiu, Jose Napolitano, Dave Russell, Iuni Margaret Laura Trist, Ruth Dooley, Craig Philip Butts - Journal: Digital Discovery - DOI: 10.1039/d6dd00189k - Keywords: graph neural network - Source URL: - Dashboard article: Abstract: We demonstrate that graph-based neural networks are capable of molecular structure elucidation of organic molecules directly from experimentally measurable NMR spectroscopic properties. Such a system is a holy grail for... ## TorchCraft: Unified binder design by inverting an all-atom structure predictor - Source: arXiv (preprints) - Date: 2026-09-17T06:37:20Z - Categories: Property Prediction, Targets & Structures - Authors: TorchCraft Team, Yu Liu, Zhouhanyu Shen, Zhengyi Li, Xikun Huang, Jiaqi Liu, Shuxian Gao, Qilin Yu, Xiayan Qin, Yucheng Zhang, Mingchen Chen - External ID: 2609.19770v1 - Keywords: AlphaFold 3 - Source URL: - Dashboard article: - PDF: - Code: Abstract: All-atom structure predictors model diverse molecular interactions, but using their learned structural priors for binder design remains challenging. Here we present TorchCraft, a unified binder-design framework that optimizes sequence logits through a frozen all-atom predictor. Implemented in TorchFold, TorchCraft combines confidence, contact, geometric, and sequence-prior objectives within a shared optimization procedure for minibinders, framework-conditioned VHHs, cyclic peptides, and ligand-binding proteins. Using pretrained AlphaFold 3 weights, TorchCraft generated representative minibinders and VHHs with experimentally measured binding across four targets in each format, without post hoc sequence redesign. Computational benchmarks further demonstrated the framework's applicability to cyclic peptides and ligand-conditioned pocket design. TorchCraft extends predictor inversion to multiple binder formats and molecular contexts, providing a common framework for reusing all-atom structural priors in design. ## 25 Years of Cardiac Ion Channel QSAR: From hERG Dominance to Multi-Channel Modeling - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-17T00:00:00+00:00 - Categories: Property Prediction, Docking & Screening, ADMET & Safety - Authors: Mateusz Iwan, Francesca Grisoni, Anastasia Pentina, Marina Garcia de Lomana, Alessandra Roncaglioni - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c03072 - Keywords: QSAR, hERG - Source URL: - Dashboard article: Abstract: In 2013 the Comprehensive In Vitro Proarrhythmia Assay (CiPA) initiative proposed multi-channel assessment for proarrhythmic risk evaluation, yet ligand-based predictive modeling has remained largely single-channel. To quantify and explain this mismatch, we survey 155 ligand-based modeling studies on cardiac ion channels (including conventional QSAR, machine learning, and pharmacophore- and other 3D ligand-based approaches) published between 2001 and mid-2026. Across this period, 88.4% of studies target hERG as their sole ion channel endpoint, a proportion that reflects data availability, regulatory incentives, and research inertia rather than scientific necessity. Although public data for Nav1.5 and Cav1.2 were sufficient for modeling years earlier, both channels received only isolated studies until 2022. Since then, multi-channel modeling has expanded substantially, with multiple groups now addressing the primary channels (hERG, Nav1.5, and Cav1.2) jointly. In contrast, the secondary channels (Kv7.1, Kv4.3, and Kir2.1) remain largely understudied: Kv4.3 and Kir2.1 are entirely unmodeled in the surveyed literature, and models for Kv7.1 remain scarce, likely due to the sparsity and heterogeneity of public data. A retrospective analysis of the reported model performance found no discernible improvements attributable to methodological advances. Reliability-related aspects, such as applicability domain and uncertainty quantification, were also rarely addressed. The field should treat joint modeling of the primary channels as the expected standard rather than a novel contribution. Other important aspects include the development of standardized benchmarks for multi-channel evaluation, integration with physiological models, and greater emphasis on prediction reliability. For the secondary channels, progress would additionally require coordinated data release or federated learning, methods suited to data-scarce targets, and community coordination, none of which individual research groups can establish alone. ## 25 Years of Cardiac Ion Channel QSAR: From hERG Dominance to Multi-Channel Modeling - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-17T00:00:00Z - Categories: Property Prediction, ADMET & Safety - Authors: Mateusz Iwan, Francesca Grisoni, Anastasia Pentina, Marina Garcia de Lomana, A. Roncaglioni - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c03072 - External ID: 539351bac9e94da6a7504b25ae2e85733002347d - Keywords: QSAR, hERG - Source URL: - Dashboard article: Abstract: In 2013 the Comprehensive In Vitro Proarrhythmia Assay (CiPA) initiative proposed multi-channel assessment for proarrhythmic risk evaluation, yet ligand-based predictive modeling has remained largely single-channel. To quantify and explain this mismatch, we survey 155 ligand-based modeling studies on cardiac ion channels (including conventional QSAR, machine learning, and pharmacophore- and other 3D ligand-based approaches) published between 2001 and mid-2026. Across this period, 88.4% of studies target hERG as their sole ion channel endpoint, a proportion that reflects data availability, regulatory incentives, and research inertia rather than scientific necessity. Although public data for Nav1.5 and Cav1.2 were sufficient for modeling years earlier, both channels received only isolated studies until 2022. Since then, multi-channel modeling has expanded substantially, with multiple groups now addressing the primary channels (hERG, Nav1.5, and Cav1.2) jointly. In contrast, the secondary channels (Kv7.1, Kv4.3, and Kir2.1) remain largely understudied: Kv4.3 and Kir2.1 are entirely unmodeled in the surveyed literature, and models for Kv7.1 remain scarce, likely due to the sparsity and heterogeneity of public data. A retrospective analysis of the reported model performance found no discernible improvements attributable to methodological advances. Reliability-related aspects, such as applicability domain and uncertainty quantification, were also rarely addressed. The field should treat joint modeling of the primary channels as the expected standard rather than a novel contribution. Other important aspects include the development of standardized benchmarks for multi-channel evaluation, integration with physiological models, and greater emphasis on prediction reliability. For the secondary channels, progress would additionally require coordinated data release or federated learning, methods suited to data-scarce targets, and community coordination, none of which individual research groups can establish alone. ## A multicriteria decision making approach for ranking gastric cancer drugs based on topological indices and physicochemical properties - Source: Scientific Reports (journals) - Date: 2026-09-17T00:00:00+00:00 - Categories: Cheminformatics, Property Prediction - Authors: Hasnain Hayat, Sarfraz Ahmad, Atef F. Hashem, Muhammad Kamran Siddiqui, Brima Gegbe - Journal: Scientific Reports - DOI: 10.1038/s41598-026-71768-9 - Keywords: QSPR - Source URL: - Dashboard article: Abstract: Chemical graph theory offers a mathematical approach to molecular structure representation and physicochemical property investigation through descriptors based on graph theory. In this study, a QSPR model is developed for analysis of some gastric cancer drug molecules by means of topological indices, regression modeling and multi-criteria decision making (MCDM) approaches. Selected topological indices were derived from the molecular graphs and used as independent variables for modeling of selected physicochemical properties by means of inverse and cubic regression models. Statistical comparison of the two types of models showed that the cubic models exhibited a goodness-of-fit than inverse models for the investigated properties in the current dataset. Additionally, selected topological indices were used as criteria in TOPSIS and SAW to obtain a mathematical ranking without taking into account any other information but the selected structural descriptors. Such rankings refer to computational comparisons based on the selected graph theoretical criteria, and do not reflect drug effectiveness, clinical applicability, superiority and recommended treatments. The study shows how chemical graph theory, QSPR analysis, and MCDM methods have been used for computation and description-based comparison of structures of molecules. ## Advanced quantitative mapping of Alzheimer’s disease neuropathology and microglial activation in post-mortem hippocampal tissue - Source: Scientific Reports (journals) - Date: 2026-09-17T00:00:00+00:00 - Authors: Terri-Leigh Stephen, Laura Korobkova, Kenneth Nguyen, Shrey Mehta, Maricarmen Pachicano, Kymry T. Jones, Bayla Breningstall, Debra Hawes, Ryan P. Cabeen, Michael S. Bienkowski - Journal: Scientific Reports - DOI: 10.1038/s41598-026-71009-z - Source URL: - Dashboard article: Abstract: We developed a high-throughput imaging workflow to spatially map Alzheimer’s disease (AD) pathology in postmortem hippocampal and medial temporal lobe sections from 65 University of Southern California Alzheimer's Disease Research Center (USC ADRC) cases classified by low, intermediate and high levels of AD neuropathologic change (ADNC). Sections were stained for microglia (Iba1), amyloid-β (4G8), and neurofibrillary tangles (NFTs; Gallyas), and analyzed using pixel-based machine learning (Ilastik). Amyloid pathology was classified as dense, diffuse, or intracellular amyloid precursor protein (APP)-positive; microglia were categorized into ramified, rod-like, and amoeboid morphologies. Diffuse amyloid plaques increased most significantly across disease, particularly in the subiculum and dentate gyrus molecular layer, while dense plaques and intracellular APP were concentrated in entorhinal and perirhinal cortices. NFT burden was elevated in males, especially in parahippocampal cortical areas. Microglial morphology shifted with AD progression, showing reduced ramified and increased amoeboid profiles in high ADNC cases. Rod microglia in CA regions were correlated with amyloid in high ADNC and tau in low ADNC cases. Memory impairment correlated more strongly with amyloid pathology in apolipoprotein E (ApoE) ε4 non-carriers, with females showing greater amyloid burden and males more NFT-related decline. These findings reveal region- and sex-specific patterns of pathology and neuroinflammation, offering new insights into mechanisms driving cognitive decline and informing future diagnostic and therapeutic strategies. ## AHAS structural changes caused by mutations result in differences in binding affinity and resistance to two herbicides in Cyperus difformis L. - Source: Frontiers in Plant Science (journals) - Date: 2026-09-17T00:00:00Z - Categories: Docking & Screening - Authors: Xiao-Tong Guo, Xiang-Ju Li, Jing-Chao Chen, Lin-Zhi Bai, Hai-Yan Yu, H. Cui - Journal: Frontiers in Plant Science - DOI: 10.3389/fpls.2026.1932845 - External ID: 040e71d190c57a31e9d3fdc326b56fdc6ca24d1b - Keywords: molecular docking, binding affinity, enzyme - Source URL: - Dashboard article: Abstract: As a malignant weed in paddy systems, Cyperus difformis L. ( C. difformis ) has evolved resistance to acetohydroxyacid synthase (AHAS, EC 2.2.1.6), also known as acetolactate synthase (ALS), inhibiting herbicides such as pyrazosulfuron-ethyl and bensulfuron-methyl due to prolonged selection pressure. This study systematically investigated the interaction mechanisms between mutated AHAS and these herbicides through biochemical and structural approaches. In this study, purified AHAS proteins of five mutated types and the wild type were obtained via prokaryotic expression, and their sensitivity and affinity for pyrazosulfuron-ethyl and bensulfuron-methyl were determined. The structural differences in different binding of AHAS binding to the two herbicides were identified via homology modeling and molecular docking. The sensitivity determination of purified AHAS indicated that AHAS with site 197 mutations had higher GR 50 and I 50 values for pyrazosulfuron-ethyl (25.60–74.78 μM) than for bensulfuron-methyl (6.62–38.32 μM). The AHAS with a site 376 mutation had higher GR 50 and I 50 values for bensulfuron-methyl than for pyrazosulfuron-ethyl, while the GR 50 and I 50 values of the AHAS with a site 574 mutation were the highest. The affinity of purified AHAS for the substrate revealed that all the mutations increased the affinity except for the Pro-197-Arg mutation. In addition, all the AHAS mutants had greater affinity for bensulfuron-methyl than for pyrazosulfuron-ethyl. Molecular docking results revealed that AHAS formed hydrogen bonds, π–π bonds, and hydrophobic interactions with pyrazosulfuron-ethyl and bensulfuron-methyl. The total number and types of interactions with pyrazosulfuron-ethyl and benzosulfuron-methyl were obviously different between the mutant AHAS and wild-type AHAS. Therefore, in this study, different mutations in AHAS induced distinct structural alterations, which reduced the affinity of the enzyme for herbicides, consequently leading to the development of weed resistance. The difference in affinity between the same mutant and the two herbicides led to a difference in resistance to the two herbicides. The results of this study are helpful for understanding the causes of weed resistance to herbicides at the structural level. ## An Integrated Strategy for Decoding Bioactive Components and Mechanism of Jingtong Granules: Chemical Characterization, Network Analysis, Pharmacokinetic and Pharmacodynamic Studies. - Source: Biomedical chromatography : BMC (journals) - Date: 2026-09-17T00:00:00Z - Categories: Docking & Screening - Authors: Jie Liu, Ren-Hu Li, Jun-Jie Qiu, Chen Li, Xi-Xi Dou, He-Li Cheng, Yu-Juan Jiang, Bing Wang, Yue Cui, Yong Chen, Teng-Fei Xu - Journal: Biomedical chromatography : BMC - DOI: 10.1002/bmc.70620 - External ID: 4167ea682358ab0b5ad2822c2a30f34b1616d2cc - Keywords: molecular docking, Pharmacokinetic - Source URL: - Dashboard article: Abstract: This study aimed to identify bioactive constituents of Jingtong Granules (JTG), a traditional Chinese medicine for cervical spondylotic radiculopathy (CSR), characterize their in vivo exposure, and explore pharmacological targets and pathways. Chemical constituents were characterized by LC-MS with GNPS-based molecular networking, and key compounds were prioritized via molecular docking; pharmacokinetic analysis was characterized in vivo exposure, and anti-inflammatory activity was evaluated via NO production in LPS-stimulated BV2 cells. Network pharmacology with GO and KEGG enrichment identified potential targets and pathways. Seventy-three components were identified in vitro and 10 in vivo. Albiflorin, puerarin, and ginsenoside Rg1 were rapidly absorbed and reduced NO production dose-dependently. Network analysis suggested MMP9 and PTGS2 as key targets, involving TNF and PI3K-AKT signaling pathways. JTG and its exposed constituents exert anti-inflammatory effects by modulating inflammatory targets and pathways. Albiflorin, puerarin, and ginsenoside Rg1 are prioritized as candidate bioactive constituents; MMP9, PTGS2, TNF, and PI3K-AKT pathways are putative mechanistic nodes requiring further validation. ## APDB: A Large-Scale Repository of Predicted 2D and 3D Aptamer Structures - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-17T00:00:00+00:00 - Categories: Cheminformatics, Property Prediction - Authors: Kabelo P Mokgopa, Kevin A Lobb, Tendamudzimu Tshiwawa - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c02031 - Source URL: - Dashboard article: Abstract: RNA aptamers are short, single-stranded oligonucleotides capable of folding into specific three-dimensional structures to bind diverse molecular targets with high affinity and specificity. In this study, we present a comprehensive computational analysis of a curated data set of over 150,000 RNA aptamers annotated with sequences, predicted secondary structures, molecular features, and minimum free energy (MFE) values. A total of 161 molecular descriptors were extracted, including pairing ratios, stem density, sequence complexity, and motif composition. Dimensionality reduction using Principal Component Analysis (PCA) and t-distributed Stochastic Neighbor Embedding (t-SNE) revealed distinct clustering patterns driven by MFE, GC/AU pairing composition, and secondary structure complexity. PCA indicated that aptamers with high AU content tend to occupy distinct structural space, while t-SNE highlighted concentrated regions of high GC pairing and stem density. A 3D surface plot of GC pairing, AU pairing, and MFE further demonstrated a nonlinear landscape, with folding stability changing in a bumpy, hill-like pattern as pairing ratios varied. These findings provide insights into how base-pairing composition influences aptamer folding stability and offer a framework for rational design. The resulting Aptamer Database (APDB) integrates these analyses into an interactive, scalable platform to support research in biosensing, therapeutics, and computational aptamer design (https://boron.ru.ac.za/database). ## Applicability-domain-constrained prediction of landfill-relevant odor descriptors using integrated chemical knowledge. - Source: Journal of hazardous materials (journals) - Date: 2026-09-17T00:00:00Z - Categories: LLMs & Agents - Authors: Bo-Yang Liao, Kun-Sen Lin, Qing An, Hao Ye, Peng Li, Xue-Fei Zhou, Ya-Lei Zhang - Journal: Journal of hazardous materials - DOI: 10.1016/j.jhazmat.2026.143649 - External ID: 3588b504919c2a3f069ca97067e9072cf1e78c05 - Keywords: XGBoost - Source URL: - Dashboard article: Abstract: Malodor emissions from municipal solid waste treatment facilities are complex environmental exposure mixtures, making rapid identification of odor-relevant compounds difficult using olfactometry or compound-by-compound GC-MS interpretation. Here, we developed a knowledge-guided molecular learning framework for interpretable and reliability-aware prediction of waste-treatment malodors. The framework integrates Morgan fingerprints, large language model-derived structure-odor rules, deterministic functional-group descriptors, and StructKG-derived hierarchical structural semantics. On a curated dataset of 3756 molecules, the fused representation with XGBoost achieved the best performance, with an AU-PRC of 0.437 and an AU-ROC of 0.869. To improve transferability to external chemical space, we introduced a multi-label applicability domain combining local similarity density with neighborhood label inconsistency, increasing in-domain AU-PRC to 0.545. External validation using compounds detected at the Shanghai Laogang waste-treatment site showed that malodor-related predictions increased from 66% outside the domain to 93% inside it. Model interpretation and theoretical odor concentration-weighted attribution identified reduced sulfur compounds, amines, volatile fatty acids, and reactive carbonyls as major odor-marker classes, supporting targeted monitoring and control of waste-treatment malodors. ## Benzaldehyde inhibits glutathione S-transferase activity and triggers reactive oxygen species accumulation in Meloidogyne incognita. - Source: Pest management science (journals) - Date: 2026-09-17T00:00:00Z - Categories: Docking & Screening - Authors: Shao-Hua Han, Guo-Hao Sun, Zi-Hao Yan, Kang Qiao - Journal: Pest management science - DOI: 10.1002/ps.71324 - External ID: 8ab14bea3fac7514f5bf77ab99e0ab78f1e9f736 - Keywords: Molecular docking - Source URL: - Dashboard article: Abstract: BACKGROUND Meloidogyne incognita poses a severe hazard to worldwide agricultural productivity, and emergence of multidrug resistance in this root-knot nematode has further exacerbated this challenge. Benzaldehyde, a volatile organic compound, is a promising nematicidal compound; however, its mode of action against M. incognita remains poorly elucidated. The present study aimed to systematically elucidate the nematicidal mechanism of benzaldehyde against M. incognita. RESULTS In vitro nematicidal assays demonstrated that benzaldehyde showed a potent nematicidal activity against second-stage juveniles (J2s) and markedly inhibited J2s hatching from eggs. Microscopic observations revealed that J2s treated with 100 mg/L benzaldehyde developed vacuolar structures after 72 h, with a greater abundance of vacuoles observed in the 400 mg/L treatment group. Notably, higher concentrations of benzaldehyde significantly suppressed the activity of glutathione S-transferase (GST), causing reactive oxygen species (ROS) overproduction and nematode death. Molecular docking and dynamics simulations confirmed that benzaldehyde could bind steadily to GST, thus weakening its antioxidant performance. Furthermore, pot experiments showed that benzaldehyde effectively reduced gall formation on cucumber roots caused by M. incognita. CONCLUSION This study uncovers a novel GST-targeted inhibitory mechanism of benzaldehyde, which shows promising prospects for the exploitation of eco-friendly nematicides. Meanwhile, it proposes an effective measure to curb the progressive multidrug resistance of M. incognita and alleviate severe agricultural economic losses. © 2026 Society of Chemical Industry. ## Beyond the drug-centric view: Advancing AI virtual cell platforms for environmental perturbation modelling. - Source: ALTEX (journals) - Date: 2026-09-17T00:00:00Z - Authors: Daniel Ukaegbu, Victor Curean, Andreas Bender, A. Maertens - Journal: ALTEX - DOI: 10.14573/altex.2606062 - External ID: 4cb65aa5051be3f5cc2e2a1b00b4f130e399e842 - Keywords: chemical space coverage - Source URL: - Dashboard article: Abstract: Environmental exposures contribute substantially to global morbidity and mortality, with their biological effects shaped by factors such as dose, exposure duration, life-stage timing, and cumulative or sequential exposure patterns. With rapid advances in AI based virtual cell (VC) technologies, current frameworks emphasize genet-ic and pharmacological perturbations, leaving environmental perturbations insufficiently modelled. This imbal-ance is due to fundamental structural limitations in data availability, chemical space coverage, and representa-tional design, as well as lacking standardized annotation of dose, timing and mixtures, which are core deter-minants of environmental perturbations. This review argues that incorporating environmental perturbations as a foundational component of virtual cell development is necessary to extend these models toward environ-mental and public health applications. We review the evolution of these perturbation predictive models from mathematical models to deep learning and foundation model architectures, evaluate the status of virtual cell efforts across genomic, transcriptomic, proteomic, metabolomic and phenomic modalities, and show the sys-tematic gaps that currently limit their applicability to environmental health. We outline key challenges, includ-ing data scarcity and bias, inadequate representation of environmental perturbations, limited multimodal inte-gration, and weak benchmarking practices. To address these issues, we propose the development of stand-ardized environmental perturbation datasets, integrated and standalone multimodal architectures, uncertainty-aware evaluation metrics, and regulatory-aligned benchmarking frameworks which are all geared toward ad-vancing in silico perturbation and non-animal testing, promoting the 3Rs paradigm. ## Bottom-up design of the operating window for direct urea synthesis from first-principles molecular thermodynamics - Source: ChemRxiv (preprints) - Date: 2026-09-17T00:00:00Z - Authors: Minwoo Kim, Seungtae Kim, Ji Woong Yu, Won Bo Lee - DOI: 10.26434/chemrxiv.15008967/v1 - External ID: 10.26434/chemrxiv.15008967/v1 - Keywords: free energy, force field - Source URL: - Dashboard article: Abstract: Urea, among the most widely utilized foods for crops, is mass-produced via the carbon-and energy-intensive Bosch–Meiser process. However, its operating condition is settled empirically, vulnerable to undersampling and suboptimal production, and its molecular mechanism is not well understood. Here we present a bottom-up design of the operating condition driven by first-principles molecular modeling. Using a machine learning force field trained on first-principles calculations of urea and relevant species, we explored the free energy surface of the reaction space from reactant to product across temperatures. We found a strong thermodynamic preference for acyl substitution over competing pathways, with a free energy minimum around 420 K arising from the balance between carbamic acid’s protonation ratio and solvation number, which respectively increase and decrease with temperature. And we confirmed that the optimal condition originates from near-critical behavior, showing a fixed ratio to the critical temperature below unity. The maximum-yield temperatures of simulation and experiment correspond when scaled to their critical temperatures, suggesting this combined approach for engineering the operating condition. ## CDGA: A Host-Based Alignment and Descriptor Analysis Tool for Cyclodextrin Inclusion Complexes - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-17T00:00:00+00:00 - Categories: Property Prediction, Reaction Informatics - Authors: Ewa Napiórkowska, Łukasz Szeleszczuk - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c02866 - Keywords: atom to atom mapping, atom mapping - Source URL: - Dashboard article: Abstract: Computational studies of cyclodextrin inclusion complexes are widespread; however, quantitative comparison of host–guest geometries remains difficult because native cyclodextrins are cyclic and pseudosymmetric, with chemically equivalent glucose units and no unique structural anchor for conventional atom-to-atom mapping. Consequently, RMSD values, guest orientations, and pose comparisons may depend on arbitrary atom numbering, molecular file representation, or software-specific alignment procedures rather than on genuine structural differences. To address this limitation, we developed CDGA, the Cyclodextrin Guest Analysis tool, an open-source KNIME-based tool for quantitative analysis of cyclodextrin host–guest complexes comprising CDGA Align for host-based alignment and CDGA NoAlign for analysis directly from their original coordinates; both workflows identify host and guest fragments, perform chemically meaningful atom mapping, and calculate standardized descriptors including guest depth, guest orientation, guest principal-axis angle, and heavy-atom RMSD, while an additional host module characterizes host geometry. CDGA was validated using controlled geometrical and chemical modifications and cross-docked and redocked poses generated for conformationally diverse native cyclodextrins, demonstrating robust and reproducible descriptor calculation despite differences in atom ordering, bond representation, host orientation, and molecular file generation, thereby enabling standardized comparison of cyclodextrin inclusion geometries, transparent reporting, benchmarking, and large-scale analysis of computational host–guest studies. ## CDGA: A Host-Based Alignment and Descriptor Analysis Tool for Cyclodextrin Inclusion Complexes - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-17T00:00:00Z - Categories: Reaction Informatics - Authors: E. Napiórkowska, Ł. Szeleszczuk - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c02866 - External ID: 41f4b99282177c13bf76cd6326d5c9764dd48b59 - Keywords: atom to atom mapping, atom mapping - Source URL: - Dashboard article: Abstract: Computational studies of cyclodextrin inclusion complexes are widespread; however, quantitative comparison of host–guest geometries remains difficult because native cyclodextrins are cyclic and pseudosymmetric, with chemically equivalent glucose units and no unique structural anchor for conventional atom-to-atom mapping. Consequently, RMSD values, guest orientations, and pose comparisons may depend on arbitrary atom numbering, molecular file representation, or software-specific alignment procedures rather than on genuine structural differences. To address this limitation, we developed CDGA, the Cyclodextrin Guest Analysis tool, an open-source KNIME-based tool for quantitative analysis of cyclodextrin host–guest complexes comprising CDGA Align for host-based alignment and CDGA NoAlign for analysis directly from their original coordinates; both workflows identify host and guest fragments, perform chemically meaningful atom mapping, and calculate standardized descriptors including guest depth, guest orientation, guest principal-axis angle, and heavy-atom RMSD, while an additional host module characterizes host geometry. CDGA was validated using controlled geometrical and chemical modifications and cross-docked and redocked poses generated for conformationally diverse native cyclodextrins, demonstrating robust and reproducible descriptor calculation despite differences in atom ordering, bond representation, host orientation, and molecular file generation, thereby enabling standardized comparison of cyclodextrin inclusion geometries, transparent reporting, benchmarking, and large-scale analysis of computational host–guest studies. ## Charge-Driven Fibril Recognition and Covalent Disruption of Aβ42 by Paddlewheel Diruthenium Complexes - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-17T00:00:00+00:00 - Categories: Cheminformatics - Authors: Alejandro Feito, Andrés R. Tejedor, Alberto Ocana, Aarón Terán, Antonello Merlino, Daniela Marasco, Santiago Herrero, Jorge R. Espinosa - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c02388 - Source URL: - Dashboard article: Abstract: The inhibition of Aβ42 (β-amyloid) fibril formation is a key therapeutic strategy in Alzheimer’s disease research. Paddlewheel diruthenium complexes have shown promising activity against Aβ42 aggregation and preformed fibril disaggregation, yet their molecular mode of action remains poorly understood. In this work, we perform atomistic simulations to explore how charge modulation influences the interactions of three analogous paddlewheel diruthenium complexes, the parent neutral complex \[Ru2Cl(D-p-FPhF)(O2CCH3)3\], and its anionic \[Ru2Cl2(D-p-FPhF)(O2CCH3)3\]− and cationic \[Ru2(D-p-FPhF)(O2CCH3)3\]+ counterparts (D-p-FPhF– is the N,N′-bis(4-fluorophenyl)formamidinato ligand) with Aβ42. Our results indicate that electrostatic tuning governs binding affinity and the extent of interaction across the Aβ42 fibril surface. As the complexes’ charge changes from −1 to +1, the interaction pattern shifts from localized contacts to widespread, multi-site engagement encompassing key charged, aromatic, and hydrophobic regions of Aβ42. This enhanced binding correlates with longer-lived, thermodynamically stable interactions at the fibril interface, which effectively lower the free energy penalty for fibril disassembly. Overall, our findings propose a mechanism in which charge-dependent activation through ligand exchange enhances fibril recognition and promotes disruptive binding modes, demonstrating the potential of charge-tunable diruthenium complexes as therapeutic modulators of Aβ42 fibril stability. ## Chemistry-Informed Multimodal Model for Structure Elucidation in Automated Reaction Discovery - Source: ChemRxiv (preprints) - Date: 2026-09-17T00:00:00Z - Categories: Cheminformatics - Authors: Maik G. Niedziella, Philipp M. Pflüger, Ajnabiul Hoque, Frank Glorius - DOI: 10.26434/chemrxiv.15008991/v1 - External ID: 10.26434/chemrxiv.15008991/v1 - Keywords: SMILES, transformer - Source URL: - Dashboard article: Abstract: Rapid and reliable structure elucidation remains a central bottleneck in automated reaction discovery, where highthroughput experimentation can generate large numbers of crude reaction mixtures faster than they can be interpreted. Gas chromatography coupled with mass spectrometry (GC-MS) provides rapid and information-rich analytical data, but identification of unknown reaction products still relies heavily on expert interpretation or spectral library matching, limiting applicability to novel compounds. Here we introduce Dynamic Mass-Aware Identification from Reaction Knowledge (DynaMAIK), a multimodal transformer framework for reaction-aware structure elucidation from GC-MS data. DynaMAIK combines electron ionization mass spectra with reaction context encoded as reactant and reagent SMILES to predict product structures and molecular formulas. Trained on more than three million reaction–spectrum pairs derived from curated reaction databases, spectral simulation and experimental spectra, DynaMAIK achieved 87% top-1 and 94% top-10 structure accuracy on a held-out experimental test set. Comparisons with spectrum-only and reaction-only models, together with ablation and scrambling experiments, show that spectral evidence and reaction-derived context provide complementary constraints for product identification. Application to high-throughput thiolation reactions demonstrates accurate product assignment directly from experimental screening data. DynaMAIK establishes context-aware GC-MS interpretation as a scalable route toward automated analytical feedback in reaction discovery workflows. ## Chromatographic Profiling of Encorafenib Impurities Using HPLC and LC-MS/MS Structural Characterization and In Silico Toxicity Evaluation of Degradation Products - Source: Journal of Applied Pharmaceutical Science (journals) - Date: 2026-09-17T00:00:00Z - Categories: ADMET & Safety - Authors: K. Babu, Gowrisankar Reddipalli - Journal: Journal of Applied Pharmaceutical Science - DOI: 10.1177/22313354261472049 - External ID: a97d659f8ae8672d13122d7fed3fc3b10caec14c - Keywords: ADMET prediction, hERG, ADMET - Source URL: - Dashboard article: Abstract: This study systematically evaluated the impurities of Encorafenib along with structural elucidation, and in silico toxicological potential of its degradation products (DPs). The HPLC resolution of Encorafenib and its impurities was achieved on a Waters Symmetry C18 column using an isocratic mobile phase of methanol and 0.1 M ammonium acetate (pH 4.1, 60:40 v/v) at a flow rate of 0.85 mL/min with detection at 231 nm. The method effectively resolves Encorafenib and its three known impurities (Rs > 2) with acceptable ICH validation criteria such as linearity ( r ² > 0.999), sensitivity (LOD: 0.075 µg/mL), precision (RSD < 2%), robustness, and specificity. The base hydrolysis induces 16.31% degradation of Encorafenib with the formation of three new DPs, whereas acid stress produces 8.28% degradation. The LC-MS/MS analysis with fragmentation pattern was utilized for structural elucidation of major DPs. The in silico toxicity assessment proves that acid DP 2 and base DPs 3 and 4 display moderate toxicity (Class 4; LD 50 : 1000–1300 mg/kg), whereas base DP 1 shows high toxicity (Class 3; LD 50 : 155 mg/kg). The ADMET prediction using the pkCSM tool indicates favorable absorption characteristics and absence of predicted mutagenicity. Hepatotoxicity was predicted for all DPs; DP3 shows hERG II inhibition, and DP4 displays a broad toxicity profile. The developed method and comprehensive characterization of DPs provide a solid foundation for routine analysis and stability studies of Encorafenib. ## Comparative study of ensemble-based uncertainty quantification methods for neural network interatomic potentials - Source: Machine Learning: Science and Technology (journals) - Date: 2026-09-17T00:00:00+00:00 - Authors: Yonatan Kurniawan, Mingjian Wen, Ellad B Tadmor, Mark K Transtrum - Journal: Machine Learning: Science and Technology - DOI: 10.1088/2632-2153/ae9fb4 - Keywords: neural network potentials, ab initio - Source URL: - Dashboard article: Abstract: Machine learning interatomic potentials (MLIPs) enable atomistic simulations with near first-principles accuracy at substantially reduced computational cost, making them powerful tools for large-scale materials modeling. The accuracy of MLIPs is typically validated on a held-out dataset of ab initio energies and atomic forces. However, accuracy on these small-scale properties does not guarantee reliability for emergent, system-level behavior—precisely the regime where atomistic simulations are most needed, but for which direct validation is often computationally prohibitive. As a practical heuristic, predictive precision—quantified as inverse uncertainty—is commonly used as a proxy for accuracy, but its reliability remains poorly understood, particularly for system-level predictions. In this work, we systematically assess the relationship between predictive precision and accuracy in both in-distribution (ID) and out-of-distribution (OOD) regimes, focusing on ensemble-based uncertainty quantification (UQ) methods for neural network potentials, including bootstrap, dropout, random initialization, and snapshot ensembles. We use held-out cross-validation for ID assessment and calculate cold curve energies and phonon dispersion relations for OOD testing. These evaluations are performed across various carbon allotropes as representative test systems. We find that uncertainty estimates can behave counterintuitively in OOD settings, often plateauing or even decreasing as predictive errors grow. These results highlight fundamental limitations of current UQ approaches and underscore the need for caution when using predictive precision as a stand-in for accuracy in large-scale, extrapolative applications. ## Conditional deep generative modeling of blood-based infrared spectra enables controlled in-silico phenotyping studies - Source: npj Digital Medicine (journals) - Date: 2026-09-17T00:00:00Z - Authors: N. Leopold-Kerschbaumer, Timo Halenke, Selina Süzeroğlu, Moritz Jung, Mariia Seleznova, Nicole Thorisch, Jeanette M. Lorenz, T. Bocklitz, B. Eskofier, Gitta Kutyniok, F. Krausz, K. Kepesidis - Journal: npj Digital Medicine - DOI: 10.1038/s41746-026-03226-9 - External ID: 075f9d34acfe0582d84f1a2c956b0b264cff585b - Keywords: Diffusion Model, GAN, Variational Autoencoder - Source URL: - Dashboard article: Abstract: Infrared molecular fingerprinting of blood offers a scalable, minimally invasive window into human physiology, but limited follow-up, imbalanced cohorts, and restricted access to diverse phenotypes constrain systematic studies. Here, we introduce a conditional deep generative framework for synthesizing blood-based infrared spectra that preserves individual-level structure while allowing controlled manipulation of demographic and anthropometric covariates. Using 25,308 spectra from 5,863 ostensibly healthy participants in the longitudinal Health4Hungary - Hungary4Health cohort, we train a Conditional Variational Autoencoder, a Conditional Boundary Equilibrium GAN, and a Conditional Diffusion Model to generate blood-based infrared spectra conditioned on age, sex, and body mass index. We show that the generated spectra closely match held-out real data across multiple levels and faithfully encode demographic and anthropometric information. We further demonstrate two in-silico applications: modeling of individualized healthy aging trajectories that follow cohort-level aging manifolds while retaining individual-specific characteristics, and targeted augmentation of underrepresented body mass index categories. Together, these results demonstrate the feasibility of utilizing conditional generative modeling of blood-based infrared spectra for virtual cohort construction, cohort balancing, and controlled in-silico phenotyping, paving the way toward more comprehensive and data-efficient studies in precision health. ## CUE: A Chemical Uncertainty-Aware Embedding Framework for Multimodal Drug Selectivity Prediction - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-17T00:00:00+00:00 - Categories: Cheminformatics, Property Prediction, Docking & Screening, Reaction Informatics - Authors: Jin Hyuk Kim, Gyeong Hwan Kim, Hyeon Jun Park, Jonghwan Choi - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c01761 - Keywords: QSAR, virtual screening, molecular fingerprints - Source URL: - Dashboard article: - Code: Abstract: Accurate prediction of drug selectivity is critical for prioritizing candidate compounds with reduced off-target effects in AI-based drug discovery. Although selectivity can be inferred indirectly from drug–target affinity (DTA) prediction, cumulative errors from affinity predictions across multiple targets can reduce reliability, motivating the development of methods that directly predict compound-level selectivity. We propose a Chemical Uncertainty-aware Embedding (CUE) framework that integrates molecular fingerprints and 2D molecular structure image embeddings. The two modalities are combined via Loss Trajectory Analysis for Uncertainty (LTAU)-based weighted fusion, which adaptively reweights features according to their predictive reliability. Across eight benchmark data sets, CUE achieved RMSE values ranging from 0.163 to 1.691 and outperformed affinity-based and quantitative structure–activity relationship (QSAR) baseline models. Furthermore, a virtual screening case study for EGFR(T790M/C797S) inhibitors further demonstrated its utility for identifying selective hits. The source code is available at https://github.com/jjjabcd/CUE. ## Curcumin Alleviates Endometriosis in Sexually Mature Female Mice by Targeting PKA and Inhibiting De Novo Estrogen Synthesis - Source: Biomedicines (journals) - Date: 2026-09-17T00:00:00Z - Categories: Docking & Screening - Authors: Hai-Yang Zhao, Xu-Dong Chen, Si-Rui Zhang, Jie-Yi Zhang, Yue-Yuan Wang, Zheng-Rong Xia, Hui Wang, Zhao-Hui Xu - Journal: Biomedicines - DOI: 10.3390/biomedicines14092099 - External ID: 188e9b57812a192c158225dc919f5ff92b9313ea - Keywords: molecular docking, receptor, kinase - Source URL: - Dashboard article: Abstract: Background: Endometriosis is an estrogen-dependent disorder with limited treatment options. The precise molecular targets of curcumin and its effects on local estrogen synthesis in this disease remain unclear. Methods: We investigated curcumin’s efficacy and mechanism via network pharmacology, in vitro functional assays in ectopic endometrial cells, an allogeneic mouse model, and target validation through molecular docking, thermal shift, and hydrolysis stability assays. Results: Curcumin significantly inhibited ectopic cell malignant phenotypes, reducing migration distance by 55.60 ± 7.32%, 80.34 ± 11.19%, and 82.88 ± 13.67% at 5, 10, and 15 μM, respectively, and decreasing invasive cell numbers by 40.38 ± 7.67%, 58.33 ± 12.70%, and 60.18 ± 9.07%, respectively (all p < 0.05). In a mouse model, curcumin (200 mg/kg/day) reduced lesion number by 38.46% (2.167 ± 0.752 vs. 1.333 ± 0.516, p = 0.0493) and lesion volume by 59.73% (0.0497 ± 0.0173 vs. 0.0200 ± 0.0130 mm3, p = 0.00732). It also suppressed de novo estrogen synthesis and modulated the estrogen receptor α (ERα) and β (ERβ) ratio. Mechanistically, curcumin directly bound to and inhibited protein kinase A (PKA), a key kinase in the cAMP signaling pathway. Conclusions: We first demonstrate that curcumin alleviates endometriosis by targeting PKA to inhibit local estrogen synthesis, identifying PKA as a novel therapeutic target and curcumin as a promising treatment candidate. ## Data-Driven Exploration of Literature-Derived Catalyst and Reaction Spaces for the Vapor-Phase Aldol Condensation of Acetate with Formaldehyde - Source: ChemRxiv (preprints) - Date: 2026-09-17T00:00:00Z - Authors: Simon B. Verstraeten, Yayati Naresh Palai, Ekaterina V. Makshina, Bert F. Sels - DOI: 10.26434/chemrxiv.15008977/v1 - External ID: 10.26434/chemrxiv.15008977/v1 - Keywords: reaction conditions - Source URL: - Dashboard article: Abstract: Emerging catalytic processes are often characterized by sparse, heterogeneous, literature-derived datasets, limiting the applicability of conventional predictive machine learning. Herein, we demonstrate a robust approach for extracting mechanistic insights from such challenging datasets using interpretable machine learning. A database comprising 375 observations from 71 publications on the vapor-phase thermocatalytic aldol condensation of acetate with formaldehyde to acrylate was compiled, including catalyst composition, reaction conditions, and catalytic performance. Unsupervised clustering identifies three catalyst families: titanium and vanadium phosphorus oxides, alumina-supported Cs- and Ba-based catalysts, and silica-supported Cs-based catalysts. The small dataset size, restricted catalyst diversity, and inter-study variability result in substantial model overfitting and constrain the accurate prediction of catalyst performance. Robust model development relies on regularized tree-based estimators, publication-grouped cross-validation, and sensitivity analysis across model architectures and random initializations. SHAP analysis identifies liquid hourly space velocity and reactant molar ratio as the dominant operational variables governing acrylate productivity. The presence of water, methanol, or ester-based acetate precursors lowers productivity, while reaction temperature exhibits limited influence. Metal phosphorus oxide and silica-supported Cs materials are ranked as the most and least active catalyst families, respectively. Despite catalyst deactivation being the major limitation, time-on-stream profiles remain largely underreported. Correlation analysis associates the deactivation of metal phosphorus oxide and alumina-supported catalysts primarily with acrylate productivity, consistent with acrylate polymerization, whereas silica-supported Cs catalysts exhibit greater variability, suggesting additional deactivation pathways. Overall, this work provides a transferable framework for extracting reliable insights from sparse literature-derived datasets using interpretable machine learning and robust validation. ## Diff-SelfA-PepGSI: A Diffusion-Based Framework for De Novo Generation, Screening, and Morphological Identification of Self-Assembled Peptide Sequences - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-17T00:00:00+00:00 - Categories: Property Prediction, Design de novo, Targets & Structures - Authors: Weizhong Sun, Yuning Ma, Xiumin Shi, Mingshan Wei, Ao He, Weizhi Wang - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c02911 - Keywords: de novo design, generative model - Source URL: - Dashboard article: Abstract: Peptide self-assembly is a ubiquitous molecular phenomenon in which short amino acid sequences spontaneously organize into diverse nanostructures. These assemblies exhibit distinct morphologies closely linked to their biological functions and potential applications, making them attractive candidates for biomaterials in drug delivery, regenerative medicine, and diagnostics. However, designing and predicting peptide sequences with specific self-assembled morphologies remains highly challenging due to the complex interplay of noncovalent interactions, including hydrogen bonding, hydrophobic effects, and π–π stacking. Although deep learning has enabled accurate prediction of peptide self-assembly propensity, rational de novo design of peptides with targeted assembly behavior remains largely unexplored. In this study, we propose Diff-SelfA-PepGSI, a novel pipeline for the morphology-specific design of self-assembling peptides. This three-stage framework comprises: (i) generation, in which a diffusion-based generative model produces candidate self-assembling peptide sequences; (ii) screening, in which pretrained protein language models extract deep semantic features and a deep discriminative network performs sequence optimization and property screening; and (iii) identification, in which a dual-branch structure extracts features and a dedicated classifier identifies the specific nanostructures formed. By integrating sequence generation, predictive screening, and morphology identification, this workflow establishes a proof-of-concept de novo design pipeline that links peptide sequences directly to their target morphologies. Diff-SelfA-PepGSI demonstrates superior performance compared to state-of-the-art models in both screening and identification tasks. Experimental validation preliminarily demonstrates the feasibility of the de novo design pipeline, successfully generating self-assembling peptides such as LAPFLA. Under the reported experimental conditions, 4 of 11 generated candidates (36.4%) exhibited the target morphology predicted by the model. Within the challenging context of de novo morphology-specific peptide design, this success rate supports the potential of Diff-SelfA-PepGSI as an early stage self-assembling peptide generation platform. ## Diff-SelfA-PepGSI: A Diffusion-Based Framework for De Novo Generation, Screening, and Morphological Identification of Self-Assembled Peptide Sequences - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-17T00:00:00Z - Authors: Wei-Zhong Sun, Yu-Ning Ma, Xiu-Min Shi, Ming-Shan Wei, Ao He, Weizhi Wang - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c02911 - External ID: 5108845b0422823da579a2bea0ab6f3f3c4eb8df - Keywords: de novo design, generative model - Source URL: - Dashboard article: Abstract: Peptide self-assembly is a ubiquitous molecular phenomenon in which short amino acid sequences spontaneously organize into diverse nanostructures. These assemblies exhibit distinct morphologies closely linked to their biological functions and potential applications, making them attractive candidates for biomaterials in drug delivery, regenerative medicine, and diagnostics. However, designing and predicting peptide sequences with specific self-assembled morphologies remains highly challenging due to the complex interplay of noncovalent interactions, including hydrogen bonding, hydrophobic effects, and π–π stacking. Although deep learning has enabled accurate prediction of peptide self-assembly propensity, rational de novo design of peptides with targeted assembly behavior remains largely unexplored. In this study, we propose Diff-SelfA-PepGSI, a novel pipeline for the morphology-specific design of self-assembling peptides. This three-stage framework comprises: (i) generation, in which a diffusion-based generative model produces candidate self-assembling peptide sequences; (ii) screening, in which pretrained protein language models extract deep semantic features and a deep discriminative network performs sequence optimization and property screening; and (iii) identification, in which a dual-branch structure extracts features and a dedicated classifier identifies the specific nanostructures formed. By integrating sequence generation, predictive screening, and morphology identification, this workflow establishes a proof-of-concept de novo design pipeline that links peptide sequences directly to their target morphologies. Diff-SelfA-PepGSI demonstrates superior performance compared to state-of-the-art models in both screening and identification tasks. Experimental validation preliminarily demonstrates the feasibility of the de novo design pipeline, successfully generating self-assembling peptides such as LAPFLA. Under the reported experimental conditions, 4 of 11 generated candidates (36.4%) exhibited the target morphology predicted by the model. Within the challenging context of de novo morphology-specific peptide design, this success rate supports the potential of Diff-SelfA-PepGSI as an early stage self-assembling peptide generation platform. ## Dry Selection and Wet Evaluation of New 1,2-Disubstituted Nitroindazolin-3-One Derivatives as Promising Agents Against Trypanosoma cruzi - Source: Pharmaceuticals (journals) - Date: 2026-09-17T00:00:00Z - Authors: J. Castillo-Garit, Josué Pozo-Martínez, Esteban Rocha-Valderrama, Cristian Rojas-Peña, Karen Acosta-Quiroga, V. Arán, Claudio Olea-Azar, Gerardo M. Casañola-Martín, B. Rasulev, F. Torrens, F. Pérez-Giménez, Mauricio Moncada-Basualto - Journal: Pharmaceuticals - DOI: 10.3390/ph19091473 - External ID: 1acf5472370f48490f6c1d2584c5e1ee18552614 - Keywords: random forest, molecular descriptors, IC50 - Source URL: - Dashboard article: Abstract: Background/Objectives: Chagas disease is endemic to 21 Latin American countries and is a great public health problem. Current chemotherapy remains unsatisfactory; consequently, the need to search for new drugs persists. The aim of this work is to develop a machine learning computational model, which allows the identification of new chemical compounds with potential trypanosomicidal activity. Methods: A large dataset of 584 compounds, obtained from the Drugs for Neglected Diseases initiative, is used to develop the computational model. AlvaDesc v3.0.14 software is used to calculate the molecular descriptors, and Scikit-learn of Python to obtain the random forest. Results: The best random forest model shows accuracy of 82.1% for the training set and near to 79% for the test set, achieving specificity values over 84.9%, and the false alarm rate values were almost under 15% for both sets. As an experiment of virtual lead generation, the present model is finally satisfactorily applied to the virtual evaluation of a series of 1,2-disubstituted nitroindazolin-3-ones obtained a good agreement between the predicted activity and the experimental assays performed. Compounds 1c and 2c stood out as the most active in the series, with half-maximal inhibitory concentration (IC50) values of 71.3 and 40.9 μM, respectively. Mechanistic analyses suggest that the presence of the nitro group may promote the generation of reactive oxygen species through enzymatic redox activation, potentially involving T. cruzi nitroreductases (TcNTRs), leading to oxidative-stress-mediated parasite death. For the most active compounds, the data are consistent with intracellular hydroxyl radical generation through enzymatic redox processes. Conclusions: Even though none of them resulted more active than nifurtimox, the current results constitute a step forward in the search for efficient ways to discover new lead antitrypanosomals. ## Evaluating a Lightweight Neural Network for Aggregate Isotope Distribution Prediction - Source: ChemRxiv (preprints) - Date: 2026-09-17T00:00:00Z - Authors: John G. Pavek, Josiah Grimes, Brian L. Frey, Nathan V. Welham, Lloyd M. Smith, Michael T. Marty - DOI: 10.26434/chemrxiv.15006709/v2 - External ID: 10.26434/chemrxiv.15006709/v2 - Source URL: - Dashboard article: Abstract: Isotope distribution prediction is an important part of mass spectrometry data analysis. A variety of strategies have been developed, including brute-force polynomial methods and Fourier-transform (FT) convolution methods. Here, we present a novel neural network (NN) approach to isotope prediction. These NN-based tools are distributed in a new package, IsoGen, alongside FT-based tools. We show that NNs perform as well as existing approaches when predicting isotope distributions for molecules with natural isotope ratios. We then demonstrate the capability of NNs for transfer learning, a method by which existing models that have been trained to perform one task can be reused as a starting point to train models to perform a second task. Here, a NN that has been trained to predict isotope distributions for molecules with natural isotope ratios can be retrained to predict distributions for molecules with perturbed isotope ratios. We also demonstrate that these training distributions do not need to be produced theoretically but may be extracted from experimental data where the underlying isotopic composition is not known. Overall, NNs provide a robust method for isotope prediction that can be extended to applications where the isotope distributions can be measured but not easily predefined. ## Explainable Artificial Intelligence to Unveil Patterns of Antioxidant Peptides for Free Radical Regulation - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-17T00:00:00+00:00 - Categories: Cheminformatics - Authors: Maria Carolina J. A. Schneider, Bárbara Saraiva Souza, Leonardo Vasconcelos Ferreira, Marina Geisiely Damaso, André Silva Pimentel - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c01872 - Source URL: - Dashboard article: Abstract: Antioxidant peptides are emerging as promising therapeutic and nutraceutical agents because of their capacity to neutralize free radicals, modulate oxidative stress, and contribute to disease prevention in fields ranging from oncology to dermatology and food science. However, traditional computational approaches often fail to capture the biochemical semantics and evolutionary context of peptide sequences, limiting the predictive accuracy and mechanistic interpretability. Here, we present an explainable deep learning framework integrating evolutionary scale modeling (ESM) embeddings with a temporal convolutional network (TCN) and long short-term memory (LSTM) architecture to classify antioxidant peptides and uncover key sequence determinants of activity. ESM embeddings provide rich evolutionary and structural priors, enabling the model to capture both local k-mer motifs and global sequence dependencies. To ensure transparency, we employed a multimethod interpretability strategy using Anchor, LIME, and SHAP, which identified robust, potentially biologically meaningful hypotheses containing residues rich in cysteine (C), arginine (R), lysine (K), histidine (H), aspartic acid (D), glutamic acid (E), threonine (T), glutamine (Q), proline (P), methionine (M), tyrosine (Y), and tryptophan (W) as central to antioxidant activity. Anchor offered deterministic, high-precision motif rules, while LIME and SHAP highlighted redundant and context-dependent contributions, together forming a layered interpretability framework. This study not only achieves state-of-the-art classification performance but also bridges black-box modeling and biochemical understanding, offering a generalizable, hypothesis-generating platform for rational peptide design and discovery that is important in human health. ## Fish Epigenetics: Molecular Mechanisms, Environmental Adaptation, and Emerging Computational Approaches - Source: Oceans (journals) - Date: 2026-09-17T00:00:00Z - Authors: M. H. Molla, M. Abualreesh, Mohammad Saeed Aljazza Alqahtani, A. Haridi, Mohammed F. Khayat, Bushra Jahan, Md. Shafiqul Islam - Journal: Oceans - DOI: 10.3390/oceans7050079 - External ID: 2385685d857c5e1ae6209b3d9f4dc374a2f2e721 - Source URL: - Dashboard article: Abstract: Epigenetic regulation has transformed our understanding of how fish adapt to changing environments by modulating gene expression without altering the underlying DNA sequence. This review explores the “dark mastery” of fish epigenetics by providing mechanistic insights into the principal epigenetic processes, including DNA methylation, histone modifications, chromatin remodeling, and non-coding RNAs, that govern development, immunity, stress responses, and disease susceptibility. These regulatory mechanisms enable fish to respond dynamically to environmental stressors such as temperature fluctuations, salinity shifts, hypoxia, pollutants, ultraviolet radiation, and nutritional changes, thereby influencing physiological resilience, reproductive performance, and survival. Recent advances in next-generation sequencing and multi-omics technologies have substantially expanded our understanding of the fish epigenome, while bioinformatics has become indispensable for integrating and interpreting complex genomic, transcriptomic, and epigenomic datasets. Furthermore, artificial intelligence (AI) and machine learning (ML) are emerging as powerful approaches for biomarker discovery, predictive modeling of disease susceptibility, environmental risk assessment, and precision aquaculture. The integration of epigenetics with bioinformatics and AI provides unprecedented opportunities to decipher complex regulatory networks, identify adaptive epigenetic signatures, and develop data-driven strategies for improving fish health and aquaculture sustainability. Despite these advances, important challenges remain, including limited species-specific epigenomic resources, difficulties in multi-omics integration, model interpretability, and the need for standardized analytical frameworks. This review highlights current knowledge, emerging computational approaches, and future perspectives for translating epigenetic discoveries into sustainable aquaculture practices and aquatic ecosystem conservation under accelerating environmental change. ## From Computational Chemistry to Generative Models: A Survey of AI-Driven Small-Molecule Drug Discovery - Source: ChemRxiv (preprints) - Date: 2026-09-17T00:00:00Z - Categories: Cheminformatics, Property Prediction, Docking & Screening, ADMET & Safety, Design de novo - Authors: Houman Kazemzadeh, Kiarash Mokhtari, Seyed Reza Tavakoli, Nazanin Mirzaei, Ali Reza Keivanimehr, Shayan Majidifar, Ali Behrad, Farbod Davoodi, Sina Khosravi, Behdad Alikhani, Ali Sabzi, Parham Abed Azad, Arshia Abolghasemi, Afshin Cheraghzade, Siavash Ahmadi, Babak Khalaj, Mohammad Hossein Rohban, Gholamali Aminian, Ali Baheri, Koorosh Aslansefat - DOI: 10.26434/chemrxiv.15008163/v3 - External ID: 10.26434/chemrxiv.15008163/v3 - Keywords: QSAR, de novo design, SMILES, SELFIES, pharmacophore modeling, reinforcement learning, Generative Models, equivariant, graph neural network, diffusion models, molecular representations, pharmacokinetic - Source URL: - Dashboard article: Abstract: Small-molecule drug discovery requires navigating enormous chemical space to identify candidates that are simultaneously potent, selective, synthetically accessible, and acceptable across pharmacokinetic and safety profiles; modern generative AI extends a long computational medicinal chemistry lineage that began with QSAR, pharmacophore modeling, and early structure-based de novo design. This survey focuses primarily on developments from 2017 to 2025, with the literature updated through 31 January 2026. Using a medicinal-chemistry lens, it first establishes the shared machinery of the field: molecular representations such as SMILES, SELFIES, graphs, fingerprints, three-dimensional coordinates, voxels, and latent vectors; commonly used datasets and benchmarks; and graph neural network encoders, including message-passing, directional, and equivariant architectures. We compare five principal generative mechanisms—autoregressive models, variational autoencoders, generative adversarial networks, normalizing flows, and diffusion models—together with reinforcement learning as a major goal-directed optimization framework. Across these approaches, we show that current methods offer complementary rather than interchangeable trade-offs, and we highlight recurring challenges, including validity versus synthesizability, limited prospective experimental validation, mode collapse and diversity loss, stereochemical and physical realism in three-dimensional design, computational cost, data scarcity, and the emerging convergence toward scaffold-conditioned generation, hybrid architectures, developability-aware optimization, and molecular foundation models. ## From Silicon to Carbon: Experimental Validation of Generative AI-Derived CNS Drug Candidates — A Systematic Review (2021–2026) - Source: ChemRxiv (preprints) - Date: 2026-09-17T00:00:00Z - Categories: Design de novo - Authors: Jad Awwad, Gul e Sehar, Fiona Benny, Obert Mutumba, Sukanya Mondal, Esha Shahbaz, Rishit Jangam, Shusant Upreti - DOI: 10.26434/chemrxiv.15008982/v1 - External ID: 10.26434/chemrxiv.15008982/v1 - Keywords: molecular generation - Source URL: - Dashboard article: Abstract: Background Generative artificial intelligence has gained traction as an approach for designing novel CNS-relevant drug candidates, yet the extent to which computationally generated molecules undergo experimental validation remains poorly characterised. Methods A systematic review was conducted in accordance with PRISMA 2020 guidelines using a prospectively registered protocol (OSF: https://doi.org/10.17605/OSF.IO/JE2DY). PubMed, Scopus, Web of Science, arXiv, and ClinicalTrials.gov were searched to March 2026. Eligible studies applied generative AI approaches to CNS-relevant targets and reported experimental wet laboratory validation. Titles and abstracts were screened independently by three reviewers; full-text screening, extraction, and quality appraisal were performed independently in duplicate. Findings were synthesised narratively; no meta-analysis was performed. Results Six studies met the inclusion criteria from 374 identified records. All performed in vitro validation; two additionally provided in vivo evidence. Despite CNS-focused objectives, one of six experimentally validated blood–brain barrier permeability. Three studies confirmed their primary computational predictions experimentally, whereas three showed partial confirmation. Validation rates, calculated as confirmed active compounds divided by compounds tested, ranged from 26 to 100 percent; these rates reflect differing selection stringency and are not directly comparable. Publicly accessible code was unavailable in four of six studies. Conclusions Among studies reaching experimental validation, blood–brain barrier confirmation, in vivo evidence, and code availability remain the exception. Because eligibility required reported validation, these findings describe successful translations and cannot estimate failure rates; they nonetheless indicate that validation depth, rather than molecular generation, is the current constraint in this field. ## Glycosides from Cistanche tubulosa suppress intestinal ACSL4-mediated ferroptosis to ameliorate depression with intestinal dysfunction symptoms through regulating arachidonic acid metabolism - Source: BMC Complementary Medicine and Therapies (journals) - Date: 2026-09-17T00:00:00Z - Categories: Docking & Screening - Authors: Li-Rong Fan, Xin-Yi Shi, Li-Juan Zhao, Jian Liu, Xue-Ping Yang, Xiao-Bo Li, Qing-Wei Zhao - Journal: BMC Complementary Medicine and Therapies - DOI: 10.1186/s12906-026-05602-0 - External ID: 35604c84f05747be985d6c54808b682e6efcabf2 - Keywords: molecular docking, binding affinity - Source URL: - Dashboard article: Abstract: Depression is highly comorbid with intestinal dysfunction, yet antidepressant therapies are limited by suboptimal efficacy and frequent gastrointestinal adverse effects. Existing cellular and animal data support the potentially dual protective effects of total glycosides (TG) from Cistanche tubulosa on stress-triggered central depression and intestinal dysfunction symptoms. However, their efficacy and underlying mechanism against depression with intestinal dysfunction symptoms remain unconfirmed to date. A chronic restraint stress mouse model of depression with intestinal dysfunction symptoms was established. Mice were divided into five groups (control, model, fluoxetine, low/high-dose TG) and treated intragastrically for 5 weeks. Behavioral tests and intestinal function assays were performed to evaluate therapeutic effects. Histological staining, immunofluorescence, biochemical analyses, ELISA, untargeted metabolomics, molecular docking, and western blotting were used to explore the underlying mechanism. TG administration significantly ameliorated chronic stress-induced depressive-like behaviors, as evidenced by restored sucrose preference, increased locomotor activity and central exploration, and reduced immobility time. Concurrently, TG improved multi-segment intestinal dysmotility, including shortening total gut transit time, restoring colonic propulsion, and enhancing gastric emptying and intestinal transit rates. Meanwhile, TG could preserve intestinal barrier integrity by mitigating epithelial damage and restoring tight junction proteins (ZO-1, occludin) and MUC2 expression, while reducing pro-inflammatory cytokines (TNF-α, IL-1β, IL-6) in the intestine, serum and hippocampus. Furthermore, metabolomic analysis identified arachidonic acid metabolism as the key altered pathway, in which TG decreased the cecal levels of arachidonic acid and 20-hydroxy-leukotriene B4, and increased the level of dinoprost that were associated with ferroptosis. Correspondingly, TG were found to suppress intestinal oxidative stress, lipid peroxidation and Fe²⁺ accumulation, reverse chronic stress-induced ACSL4 upregulation and GPX4 downregulation, and its major constituents (e.g., echinacoside) showed strong binding affinity with ACSL4, which indicated that TG could modulate the intestinal ACSL4-mediated ferroptosis. The present study demonstrates that TG exert a novel pharmacological effect in ameliorating comorbid depressive-like behaviors and intestinal dysfunction symptoms in stressed mice, suggesting that TG act via a gut-targeted pathway by suppressing intestinal ACSL4-mediated ferroptosis through regulating arachidonic acid metabolism. This novel pharmacological mechanism highlights the potential of TG as a natural product-based intervention for comorbid depression and intestinal dysfunction symptoms. ## Host egg-associated compounds elicit behavioral responses in the egg parasitoid Telenomus remus Nixon (Hymenoptera: Scelionidae) - Source: Egyptian Journal of Biological Pest Control (journals) - Date: 2026-09-17T00:00:00Z - Categories: Docking & Screening - Authors: Rui-Cheng Shi, Zhen-Xiao Li, Ping-Ting Wang, Ling-Zhi Yang, Yan Yan, Zhan-Da Ji, Zhi-Zhi Wang, Xi-Qian Ye, Xuexin Chen, Yong-Hui Xie, Pu Tang - Journal: Egyptian Journal of Biological Pest Control - DOI: 10.1186/s41938-026-00908-0 - External ID: 473dca235729188b58f96a6d2aceab99c2f2b74e - Keywords: Molecular docking, receptor - Source URL: - Dashboard article: Abstract: The egg parasitoid Telenomus remus Nixon (Hymenoptera: Scelionidae) is a promising biological control agent against lepidopteran pests. Its efficacy depends on its ability to locate and recognize host eggs, however, the egg surface chemical cues underlying host recognition by T. remus remain poorly understood. In the present study, the behavioral responses of T. remus to egg-associated compounds from three host species, Corcyra cephalonica, Spodoptera litura, and S. frugiperda using a Y-tube olfactometer were examined. GC-MS profiling revealed distinct egg-surface chemical profiles among the three host species. Behavioral assays with synthetic compounds identified seven compounds that elicited significant responses in T. remus, including 1,3-di-tert-butylbenzene, toluene, oleic acid, hentriacontane, pentacosane, cyclohexanone, and tetradeca-1,13-diene. Genome-wide searches identified 14 candidate odorant-binding protein genes (TremOBPs) and 44 candidate odorant receptor genes (TremORs), which were distributed across multiple hymenopteran OBP and OR clades. Molecular docking further predicted relatively low docking scores for 1,3-di-tert-butylbenzene with several selected TremOBPs and TremORs. These results indicated that the host egg-associated compounds elicit distinct behavioral responses in T. remus. The behaviorally active compounds identified in this study provide candidate semiochemicals for future evaluation in parasitoid host-location enhancement and biological control applications, while the candidate OBPs and ORs provide molecular targets for subsequent functional validation. ## In-Context Molecular Property Prediction with LLMs: A Blinding Study on Memorization and Knowledge Conflicts - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-17T00:00:00+00:00 - Categories: Cheminformatics, Property Prediction, LLMs & Agents - Authors: Matthias Busch, Marius Tacke, Sviatlana V. Lamaka, Mikhail L. Zheludkevich, Christian J. Cyron, Christian Feiler, Roland C. Aydin - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c00961 - Keywords: MoleculeNet, LLMs, LLM, Property Prediction, GPT, molecular property - Source URL: - Dashboard article: Abstract: The capabilities of large language models (LLMs) have expanded beyond natural language processing to scientific prediction tasks, including molecular property prediction. However, their effectiveness in in-context learning remains ambiguous, particularly given the potential for training data contamination in widely used benchmarks. This paper investigates whether LLMs perform genuine in-context regression on molecular properties or instead rely on verbatim retrieval of memorized target values. Furthermore, we analyze the interplay between pretrained knowledge and in-context information through a series of progressively blinded experiments. We evaluate nine LLM variants across three families (GPT-4.1, GPT-5, Gemini 2.5) on three MoleculeNet data sets (Delaney solubility, Lipophilicity, QM7 atomization energy) using a systematic blinding approach that iteratively reduces available information, complemented by 0-, 60-, and 1000-shot in-context sample sizes as an additional control for information access. To validate the memorization analysis and the blinding experiments, we add a positive and a negative control for the memorization experiments and structural reference baselines for the multishot experiments, as well as bootstrap confidence intervals for all results. We find no evidence of verbatim retrieval on the legacy benchmarks and show that blinding exposes conflicts between pretrained knowledge and in-context information. This work provides a principled framework for evaluating molecular property prediction under controlled information access. ## Investigation of New Cocrystal Benzimidazole 4-amino Benzoic Acid Through Structural, Spectroscopic, Thermodynamics, Topological Analysis and Molecular Docking Against Tuberculosis - Source: Chemistry Africa (journals) - Date: 2026-09-17T00:00:00Z - Categories: Docking & Screening - Authors: Saraswathi Shiva, Azar Zochedh, Jayapriyanga Karuppasamy, Asath Bahadur Sultan - Journal: Chemistry Africa - DOI: 10.1007/s42250-026-01847-x - External ID: 603a1099f09b78a5452a57f25d8e1659956fbd5d - Keywords: Molecular Docking - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Machine learning-assisted broad-spectrum aptamer discovery and intelligent detection of neonicotinoid pesticides. - Source: Biosensors & bioelectronics (journals) - Date: 2026-09-17T00:00:00Z - Categories: Docking & Screening - Authors: Ya-Lin Mo, Shi-Ping Luo, Ze-Peng Gu, Kuai-Le Wu, Ning Ma, Zhou-Ping Wang, Shi-Jia Wu, Nuo Duan - Journal: Biosensors & bioelectronics - DOI: 10.1016/j.bios.2026.119243 - External ID: aa933e4130318357a1507fa958f1b8bf409521cc - Keywords: computational screening, Molecular docking, dissociation constant - Source URL: - Dashboard article: Abstract: A machine learning-assisted computational strategy was developed for the de novo discovery of broad-spectrum aptamers targeting neonicotinoid pesticides (NEOs). Randomly generated DNA sequences were first classified according to secondary-structure similarity using t-distributed stochastic neighbor embedding (t-SNE) and K-means clustering, providing a structurally diverse initial library. Molecular docking and binding-pocket overlap analysis were then used to identify conserved interaction regions across multiple NEOs, followed by directed single-nucleotide mutagenesis and iterative computational screening to optimize broad-spectrum recognition. The resulting aptamer, NEO 3\_4, exhibited improved binding toward multiple NEOs, with a dissociation constant (Kd) of 1.48 μM. An MIL-88(Fe)-NH2@PtIr nanozyme-assisted paper-based colorimetric sensing platform was subsequently developed using NEO 3\_4. The platform achieved a linear detection range of 0.05-100 nM and a limit of detection of 26.68 pM. Combined with smartphone-based RGB extraction and artificial neural network analysis, the method enabled accurate quantification of NEOs in vegetable samples, with recoveries of 98.95-102.75% and RSDs below 3.56%. This study provides a machine learning-assisted strategy for de novo broad-spectrum aptamer discovery and a portable sensing platform for NEO monitoring. ## Molecular absorption and fluorescence property prediction via a multimodal consensus model: Winning solution for the transmittance category of the 2nd EUOS/SLAS Joint Challenge. - Source: SLAS technology (journals) - Date: 2026-09-17T00:00:00Z - Authors: Kairi Furui, Apakorn Kengkanna, Koh Sakano, Masahito Ohue - Journal: SLAS technology - DOI: 10.1016/j.slast.2026.100466 - External ID: b2cb618c32c9ad37745523c31b8e98678dd593c9 - Keywords: property prediction, MPNNs, gradient boosting, molecular representations - Source URL: - Dashboard article: Abstract: The 2nd EUOS/SLAS Joint Challenge was a competition aimed at developing reliable computational models for predicting transmittance and fluorescence properties from the chemical structures of approximately 100,000 compounds. This paper describes the method that achieved 1st place in the blind test of the Transmittance category. Our approach is based on a consensus strategy that integrates diverse predictions from multiple model types, including gradient boosting decision trees, message passing neural networks (MPNNs), and Uni-Mol2, via a weighted ensemble, covering multimodal molecular representations in 1D, 2D, and 3D. 5-fold cross-validation based on Bemis-Murcko scaffolds was employed to improve generalizability, and focal loss was adopted to handle extreme class imbalance. Post hoc comparison showed that the individual model leading under cross-validation differed from the Public-test leader in all four subtasks, whereas the final ensemble ranked between first and fifth across tasks, supporting the robustness of the consensus strategy to model selection. ## Optimization and Metabolomic Profiling of Wolffia globosa Extract Reveal Multitarget Neuroprotective Activity Against Alzheimer’s Disease with In Vitro, In Silico, and In Vivo Validation - Source: International Journal of Molecular Sciences (journals) - Date: 2026-09-17T00:00:00Z - Authors: Piya Temviriyanukul, Woorawee Inthachat, Tanongsak Laowanitwattana, Pensiri Buacheen, Uthaiwan Suttisansanee, Pornsiri Pitchakarn - Journal: International Journal of Molecular Sciences - DOI: 10.3390/ijms27188266 - External ID: c325c3eb7c3672b39c783d3b62ee9d355079f99c - Keywords: molecular docking, bioactivity, IC50 - Source URL: - Dashboard article: Abstract: Wolffia globosa, the world’s smallest flowering plant, is a sustainable protein- and phytochemical-rich food source with emerging neuroprotective potential. This study presents the first comprehensive characterization that integrates extraction optimization, multitarget bioactivity profiling, drug synergy assessment, high-resolution metabolomics, molecular docking, and in vivo validation in Drosophila melanogaster. Ethanol-based extraction was optimized using a Box–Behnken design and response surface methodology (RSM), yielding optimal conditions of 32% ethanol, 1:40 (g/mL) solid-to-solvent ratio, 60 °C, and 28 min. The optimized extract exhibited a total phenolic content (TPC) of 15.17 ± 0.27 mg GAE/g DW, a total flavonoid content (TFC) of 22.34 ± 1.71 mg QE/g DW, and an exceptional ORAC antioxidant activity of 4374.33 ± 395.65 µmol TE/g DW. Potent and selective BACE-1 inhibition was observed (IC50: 1.59 ± 0.11 mg/mL), alongside moderate AChE (IC50: 7.42 mg/mL) and BChE (IC50: 6.55 mg/mL) inhibition. Synergistic interactions with donepezil were confirmed by the Chou–Talalay combination index method, with combined AChE, BChE, and BACE-1 inhibitions reaching 60.3%, 77.0%, and 82.2%, respectively—substantially exceeding theoretical additive values. Multi-platform metabolite profiling by HPLC-QTOF-MS/MS and LC-ESI-MS/MS identified luteolin, apigenin, caffeic acid, isovitexin, schaftoside, and pinolenic acid as the principal bioactives. Molecular docking predicted favorable binding of luteolin (delta-G: −10.24 kcal/mol, AChE) and naringenin (delta-G: −8.76 kcal/mol, BACE-1) against crystal structures of human AChE (PDB: 7E3H), BChE (PDB: 4TPK), and BACE-1 (PDB: 6EQM). In vivo, the W. globosa extract significantly improved locomotor performance and suppressed brain BACE-1 activity in an Drosophila AD model over 28 days. These findings establish W. globosa as a multifunctional nutraceutical candidate for neurodegenerative disease prevention. ## Per-residue optimisation of protein structures: rapid alternative to local optimisation with constrained alpha carbons - Source: Journal of Cheminformatics (journals) - Date: 2026-09-17T00:00:00+00:00 - Categories: Cheminformatics - Authors: Ondřej Schindler, Tomáš Svoboda, Gabriela Bučeková, Radka Svobodová - Journal: Journal of Cheminformatics - DOI: 10.1186/s13321-026-01303-5 - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Per-Stage Controls and Failure Modes in a Co-Folding Virtual Screening Cascade for the Keap1–Nrf2 Interaction - Source: ChemRxiv (preprints) - Date: 2026-09-17T00:00:00Z - Categories: Library Design - Authors: Huynh-Duc Le - DOI: 10.26434/chemrxiv.15008984/v1 - External ID: 10.26434/chemrxiv.15008984/v1 - Keywords: Virtual Screening, make on demand library - Source URL: - Dashboard article: Abstract: Virtual-screening cascades assembled from co-folding and learned-affinity models are now routine, and each stage is usually trusted on the strength of its own score. We measured, stage by stage, whether those scores carry the information a practitioner would attribute to them, using the Keap1–Nrf2 protein–protein interaction and a 130,793-compound make-on-demand library as the test bed. Most measurements are negative, and none of the failure modes is specific to one of the two models tested. Two independently developed affinity models order activity-cliff pairs above chance (about 68% of 181 pairs) but carry no information about the size of the potency gap, and both couple to molecular weight; the built-in molecular-weight correction rescales without de-biasing. Neither model resolves enantiomers, including the model that generates an explicit high-confidence three-dimensional pose. Pose-model confidence is only weakly informative about geometric validity, so an explicit validity check is required rather than optional. Training-free guidance is required for stereochemical fidelity in pose generation, and diffusion steps can be halved without detectable loss. A seeded matched-decoy gate validates the pose-and-energy stage where no single physicochemical descriptor does, and the pose model reproduces all five co-crystal references below 2 Å, three of them released after its training cutoff. Single-structure MM-GBSA ordering is not validated by ensemble refinement, and the two solvation models disagree both on the ordering and on which candidates are separable at all. The control suite is released as a reusable artifact. ## Phragmites Communis (Lu Gen) Alleviates Hyperuricemia via Xanthine Oxidase Inhibition and Gut-Microbiota-Mediated Metabolite Urate Transporter Regulation - Source: Pharmaceuticals (journals) - Date: 2026-09-17T00:00:00Z - Categories: Docking & Screening - Authors: Aamir Saleem, Bushra Walayat, Qi-Di Xue, Immad Ansari, Xiao-Qing Wei, Ming Li - Journal: Pharmaceuticals - DOI: 10.3390/ph19091474 - External ID: 45e446cc64d6a31c102fa0d719189d35114b8195 - Keywords: Molecular docking - Source URL: - Dashboard article: Abstract: Background: Hyperuricemia (HU) results from excessive uric acid (UA) production/impaired excretion and is associated with gout and renal dysfunction. Gut microbiota dysbiosis may impair intestinal UA excretion, disrupt the gut barrier, and promote inflammation. Traditional Chinese medicine (TCM), with multitarget and microbiota-modulating effects, may offer a therapeutic strategy for HU. In this study, the rhizome of Phragmites communis, a TCM used for clearing heat and promoting diuresis, was investigated. Specifically, we evaluated whether P. communis could alleviate HU by suppressing xanthine oxidase (XOD) and modulating urate transporter expression, thereby reducing UA production while promoting its excretion. Method: The effects of a crude extract of P. communis were investigated with in vitro and in vivo models using molecular, 16S rRNA sequencing, and metabolomic analyses. Results: P. communis significantly decreased serum uric acid (SUA) levels by inhibiting XOD and improving renal and intestinal urate excretion, accompanied by decreased renal and increased intestinal solute carrier family 2 member 9 (SLC2A9) and enhanced ATP-binding cassette subfamily G member 2 (ABCG2) expression, restored gut barrier integrity, and suppressed inflammation. Microbiota analysis revealed enrichment of Ligilactobacillus, Prevotella, and Dubosiella. Fecal metabolomics identified hippuric acid (HA), which correlated with Ligilactobacillus abundance. In vitro, HA enhanced ABCG2, PDZ domain-containing 1 (PDZK1), and SLC2A9 expression. Molecular docking identified feruloylquinic acid (FAQ) as the favorable XOD inhibitor. Conclusions: P. communis exerts anti-hyperuricemic effects by inhibiting UA biosynthesis, promoting UA excretion, modulating gut microbiota, restoring gut barrier integrity, and suppressing inflammation, highlighting its potential as a TCM-based therapeutic intervention for HU. ## Plasma Chlorinated Paraffin Mixtures and Breast Cancer: Nonlinear Patterns in an Exploratory Case-Control Study. - Source: Environmental pollution (journals) - Date: 2026-09-17T00:00:00Z - Categories: Docking & Screening - Authors: Chun-Yan Deng, Wen-Xuan Li, Li-Qin Wang, Yu-Kun Liu, Wen-Hui Liu, Meng-Fan Ma, Bing-Yu Hou, Ya-Nan Wang, Yong Tian, Yan-Jie Zhao, Wei Gao, Yu-Xin Zheng - Journal: Environmental pollution - DOI: 10.1016/j.envpol.2026.129185 - External ID: bce5350573061da855826ef82abcb99a46a31dde - Keywords: Molecular docking, regression model - Source URL: - Dashboard article: Abstract: Increasing evidence suggests that chlorinated paraffins (CPs) exhibit estrogen-disrupting effects, thus necessitating further investigation into their impact on breast cancer development. Building upon this context, we analyzed plasma samples from 70 clinically diagnosed breast cancer patients and 101 healthy controls recruited from screening programs, aiming to quantitatively assess their internal CPs exposure levels. The results showed that short-chain chlorinated paraffins were detected in all plasma samples, with a detection rate of 100.0% and a concentration of 9.58 (4.09, 15.71) ng/mL. Medium-chain chlorinated paraffins had a detection rate of 90.1% and a concentration of 14.40 (7.83, 23.76) ng/mL. The median concentration of SCCPs were lower in the case group than in controls (6.75 vs. 11.42 ng/mL, P < 0.01), whereas MCCPs levels showed no significant difference (14.35 vs. 14.48 ng/mL, P = 0.485). Using the Bayesian kernel machine regression model, we observed an association between mixed CPs exposure and breast cancer risk-driven by C13H23Cl5-SCCPs and C16H27Cl7-MCCPs-which was evident only when all mixture components were jointly fixed at the upper exposure percentiles relative to their median levels. Further exploratory stratified analysis suggests potential associations between CPs exposure and the luminal A and luminal B subtypes of breast cancer, with C16H27Cl7-MCCPs may potentially contribute to the development of the Luminal A subtype. Molecular docking suggested the potential for interactions between selected CP congeners and estrogen receptors. As an exploratory case-control study, we found that exposure to higher concentrations of CP mixtures may be associated with breast cancer case status, providing evidence for understanding the potential health risks of CPs based on a small-sample population, and also laying the foundation for subsequent mechanistic studies and environmental health risk assessments. However, due to the limited sample size, these findings need to be confirmed in larger-scale prospective studies. ## Polymer Informatics Atlas: Property-Specific Representations, Chemistry-Aware Generalization, and Reliability-Aware Machine Learning Across Five Polymer Properties - Source: ChemRxiv (preprints) - Date: 2026-09-17T00:00:00Z - Categories: Cheminformatics - Authors: Dennis Obinna Orji - DOI: 10.26434/chemrxiv.15008969/v1 - External ID: 10.26434/chemrxiv.15008969/v1 - Keywords: RDKit - Source URL: - Dashboard article: Abstract: Polymer-property machine learning is often evaluated as if representation choice, random-split accuracy, chemical extrapolation, transfer learning, uncertainty, and chemical interpretation were a single problem. Here, these questions are separated in a six-stage Polymer Informatics Atlas built from 7,973 unique RDKit-valid polymer repeat units in the NeurIPS Open Polymer Prediction 2025 dataset. Thirty-six interpretable composition/structure descriptors and radius-4, 2,048-dimensional count Morgan fingerprints were evaluated independently and as a hybrid representation for glass-transition temperature (Tg), fractional free volume (FFV), thermal conductivity (Tc), density, and radius of gyration (Rg). Frozen property-specific models achieved untouched random-holdout R2 values of 0.608, 0.704, 0.825, 0.920, and 0.703, respectively. However, chemistry-separated testing changed this interpretation substantially: Tc fell to R2 = 0.326, whereas density remained comparatively robust at R2 = 0.874. Representation-block permutation showed that density was predominantly captured by interpretable global chemistry, while Tc and Rg depended more strongly on local/topological structure; Tg and FFV required mixed information. Masked multitask learning on the overlapping Tc-Density-Rg block used incomplete labels more efficiently than complete-case training, but gains over matched single-task neural networks were small and seed dependent. Five-member ensembles combined with splitconformal calibration produced 90% nominal intervals with empirical coverages of 95.7% (Tc), 92.3% (density), and 98.5% (Rg). Finally, 28 predictive Morgan indices were decoded to atom-centred environments after exact fingerprint-regeneration audits, and an integrated descriptor/fingerprint manifold was used for chemical-space visualization, archetype selection, and an illustrative high-Tc/low-density Pareto analysis. The results support a property-specific, domain-aware view of polymer informatics in which predictive accuracy is reported together with chemical support, uncertainty, and explicit claim boundaries rather than as a single global score. ## Predictive accuracy of peptide–target interaction models in drug discovery: a systematic review and meta-analysis - Source: Frontiers in Drug Discovery (journals) - Date: 2026-09-17T00:00:00Z - Authors: William J. Waldock, Ahmad Guni, A. Darzi, César de la Fuente-Núñez, H. Ashrafian - Journal: Frontiers in Drug Discovery - DOI: 10.3389/fddsv.2026.1894745 - External ID: 53f2c615456e2cd1067c2c0b6b1179719f71320e - Keywords: transformer, graph neural network - Source URL: - Dashboard article: Abstract: Peptides represent promising therapeutic agents due to their high specificity, biocompatibility, and capacity to modulate protein–protein interactions. However, the field faces critical challenges: inconsistent evaluation metrics, heterogeneous datasets, and poor reproducibility, which together undermine objective model comparison and benchmarking. To systematically review and quantitatively assess the predictive performance of machine learning (ML)-based peptide–target interaction (PTI) models, with emphasis on commonly reported metrics including area under the curve (AUC), concordance index (CI), and Precision. We conducted a systematic literature search across PubMed, arXiv, and Cochrane databases for studies published through 1 August 2025. Inclusion criteria required original ML-based peptide–target prediction models with quantitative performance metrics. Twenty-three studies met inclusion criteria. Fourteen reported AUC values (pooled estimate: 0.87, 95% Confidence Interval: 0.83–0.90), five reported Concordance Index values (0.90, 95% Confidence Interval: 0.89–0.91), and six reported precision (0.75, 95% Confidence Interval: 0.69–0.81). Top-performing models predominantly employed transformer or graph neural network architectures with structural input features. Critical limitations included inconsistent reporting practices, infrequent external validation, and limited data/code availability. The observation that the twenty-three studies present challenges to a pooled estimate is itself a consequence of the reporting practices we document, and is why we advance STRIDE. ML models demonstrate strong potential for PTI prediction, with leading approaches achieving robust classification and ranking performance. Nevertheless, progress is hindered by non-standardised evaluation metrics, limited transparency, and insufficient reproducibility. We introduce the STRIDE framework (encompassing Standardization, Transparency, Representativeness, Integration, Discovery, and Evidence) to establish rigorous evaluation standards, enhance methodological reproducibility, and support future evaluation of clinical applicability. ## Progress with FAK inhibitors in the patent literature (2020-present). - Source: Expert opinion on therapeutic patents (journals) - Date: 2026-09-17T00:00:00Z - Categories: Reaction Informatics - Authors: Bing-Bing Chen, Rui-Peng Feng, Jin-Bo Niu, Jian Song, Yuan-Bo Cui, Sai Zhang - Journal: Expert opinion on therapeutic patents - DOI: 10.1080/13543776.2026.2735861 - External ID: 358917ba7e794242c53b6ab00d5b71d3f7b47c3b - Keywords: SciFinder, kinase, receptor - Source URL: - Dashboard article: Abstract: INTRODUCTION Focal adhesion kinase (FAK) is a non-receptor tyrosine kinase that transduces signals from integrins, receptor tyrosine kinases, and growth factors to orchestrate cell adhesion, migration, and survival. Aberrant FAK hyperactivation promotes tumor proliferation, invasion, anti-apoptosis, and chemoresistance. Consequently, the recent U.S. FDA approval of the FAK inhibitor defactinib combined with avutometinib for recurrent KRAS-mutant low‑grade serous ovarian cancer (LGSOC) validates FAK as a clinically actionable anticancer target. AREAS COVERED This review discusses recent advances in FAK inhibitor development, focusing on small-molecule patents published from January 2020 to August 2026. A systematic search of SciFinder and WIPO databases was conducted for this period. It categorizes these novel inhibitors into several classes and summarizes their structural features, biological activities, design strategies, and structure-activity relationships (SAR). EXPERT OPINION Recent advances in FAK drug discovery have driven a transition from conventional kinase inhibition toward broader FAK pathway modulation, including improved inhibitors, dual-target strategies, and targeted protein degradation. The clinical success of defactinib-based combination therapy validates the therapeutic potential of FAK targeting, while emerging approaches offer opportunities to overcome resistance. Future progress will depend on biomarker-guided patient selection, rational combination regimens, and exploration of non-catalytic FAK functions to achieve more effective and durable therapies. ## Proteome-scale investigation of zinc-binding proteins of bread wheat - Source: Scientific Reports (journals) - Date: 2026-09-17T00:00:00+00:00 - Authors: Rajat Bhatt, Shailender Kumar Verma - Journal: Scientific Reports - DOI: 10.1038/s41598-026-64331-z - Keywords: AlphaFold3 - Source URL: - Dashboard article: Abstract: Zinc (Zn) is an essential micronutrient participating in many cellular functions and performs structural and regulatory roles in wheat. Bread wheat, one of the most widely cultivated and consumed crops worldwide, has a relatively low zinc concentration. This study focuses on the recognition of zinc-binding proteins (ZBPs) across the proteome of bread wheat ( Triticum aestivum L.) through a systematic bioinformatic analysis. The extensive size and complex nature of the wheat proteome on profiling with BLASTp produced a high volume of redundant proteins. Further, filtering was done based on metal-binding motifs of proteins by employing the MeBiPred tool, producing an output of 2,719 putative ZBPs. The putative ZBPs were modeled using AlphaFold3, and 662 high-confidence ZBPs were shortlisted for downstream analysis. These putative proteins were predicted to be predominantly present in the nucleus, followed by the chloroplast and cytoplasm. The functional analysis revealed the prevalence of domains such as peroxidase, carbonic anhydrase, and transcription regulators like C2H2 zinc finger, indicating their role in stress responses, mediating cellular signalling, and regulating nucleic acid biosynthesis. Their involvement in metabolic processes and pathways defines their importance in maintaining homeostasis, regulating gene expression, maintaining structural integrity, and balancing oxidative stress. ZBPs performing similar functions were clustered in a protein–protein interactions , revealing the network of hub genes and the presence of uncharacterised proteins regulating the core metabolic processes in plants. Identification of these uncharacterised proteins was done by cytoHubba and MCODE providing us an edge to identify ZBPs that could serve as potential biomarkers for crop biofortification programs. ## RaFT-DM: A Residue-Aware Fusion Transformer With Domain-Wise Memory for Accurate Multi-Label Protein Function Prediction. - Source: IEEE transactions on computational biology and bioinformatics (journals) - Date: 2026-09-17T00:00:00Z - Authors: Jia-Kai Zhang, Zi-Hang Zhang, Jia-Ru Yang, Wei-Ping Ding, Zhenyu Lei, Shang-Ce Gao - Journal: IEEE transactions on computational biology and bioinformatics - DOI: 10.1109/TCBBIO.2026.3734966 - External ID: edc15723335af3e64d8cd8e08ff0260562a9afff - Keywords: Transformer - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Revisiting molecular descriptors with TDiMS for interpretable intramolecular interactions based on substructure pairs - Source: Nature Computational Science (journals) - Date: 2026-09-17T00:00:00+00:00 - Authors: Lisa Hamada, Akihiro Kishimoto, Kohei Miyaguchi, Masataka Hirose, Junta Fuchiwaki, Indra Priyadarsini, Seiji Takeda - Journal: Nature Computational Science - DOI: 10.1038/s43588-026-01036-3 - Keywords: property prediction, molecular descriptors - Source URL: - Dashboard article: Abstract: Molecular descriptors play a crucial role in representing the structural features of molecules for machine learning-based physical property prediction. However, current descriptors either consider only local aspects of molecular structures or fail to effectively learn nonlocal structural features involving long-distance intramolecular interactions. Here, to address this issue, we present a descriptor named TDiMS. TDiMS effectively summarizes the enumerated pairwise topological distances between molecular substructures, thus capturing nonlocal interactions. Our evaluation shows that TDiMS successfully identifies essential features of large structures and outperforms other representative descriptors in predicting properties for which distances between substructures are a primary factor. In addition, these identified features are highly interpretable for experts in materials discovery. ## Structural Control of Phthalate Toxicity in Agricultural Soil Biofilms: Aliphatic-Chain-Length-Driven Hazards Revealed by Field Evidence, Multiomics, and Quantum Chemical Analysis - Source: Journal of Agricultural and Food Chemistry (journals) - Date: 2026-09-17T00:00:00Z - Categories: Docking & Screening - Authors: Shan-Ying He, Fei-Hong Yu, Pu Feng, Jie Chen, Zhen-Li He, Chun-Ping Yang, Ya-Ting Luo, Yu-Lu Xiang, Zhi-Heng Li - Journal: Journal of Agricultural and Food Chemistry - DOI: 10.1021/acs.jafc.6c05749 - External ID: c9953e41c9e6a833efe28fba7ee66d0cc2540f16 - Keywords: Molecular docking, density functional theory, DFT - Source URL: - Dashboard article: Abstract: Phthalates (PAEs) pose significant ecological risks in greenhouse soils because of their hydrophobicity and persistence. This study combined field survey, microcosm experiments, multiomics analysis profiling, and quantum chemical calculations to investigate how the aliphatic chain length influences the toxicity of PAEs toward soil biofilms. Results showed that longer-chain PAEs caused stronger inhibition of biofilm formation, reduced extracellular polymeric substance polysaccharide production, enhanced oxidative stress, and disrupted microbial community structure. Multiomics revealed that PAEs primarily affected energy metabolism, electron transport, and glycerophospholipid metabolism. Molecular docking and density functional theory (DFT) calculations further suggested that PAEs preferentially interact with the Trp198 region of NADH dehydrogenase I, with the binding strength increasing with the aliphatic chain length. This study establishes a structural toxicology framework linking the molecular structure to biofilm dysfunction, providing mechanistic insights into the ecological risk assessment of PAEs. ## Synthesis, Biological Evaluation, Molecular Docking, and Dynamics, DFT Studies of Amide-functionalized 1,2,3-Triazole-Acridone Conjugates - Source: Polycyclic Aromatic Compounds (journals) - Date: 2026-09-17T00:00:00Z - Categories: Docking & Screening - Authors: Krishna V. Lathi, Mininath K. Bhalmode, M. K. Lande, Rohini R. Suradkar, Dipti D. More, D. Raut, A. G. Navaneeth, Karthikeyan Subramani, Ashruba B. Danne, B. Shingate - Journal: Polycyclic Aromatic Compounds - DOI: 10.1080/10406638.2026.2729267 - External ID: 6d70db92142b3272b1c23a8a65d11ac1eead8448 - Keywords: Molecular Docking, DFT - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Synthesis, characterization and molecular docking studies of novel 6-Imino-2-Methyl-4-substituted- 6H -1,3-Thiazine-5-Carbonitrile derivatives as potential antibacterial agents - Source: Main Group Chemistry (journals) - Date: 2026-09-17T00:00:00Z - Categories: Docking & Screening - Authors: Suhas Y. Salunkhe, S. D. Patil, P. J. Menezes, S. Vartale - Journal: Main Group Chemistry - DOI: 10.1177/10241221261476810 - External ID: 0e4bf7ad6f582c5bd7332a35ceb06c282005dc1b - Keywords: molecular docking, AutoDock Vina - Source URL: - Dashboard article: Abstract: Novel 6-imino-2-methyl-4-substituted-6H-1,3-thiazine-5-carbonitrile derivatives were synthesized via cyclocondensation of thioacetamide and bis(methylthio)methylene malononitrile in DMF using K 2 CO₃. The key 4-methylthio intermediate underwent nucleophilic substitution with various anilines, phenols, heterocyclic amines, and active methylene compounds to produce a diverse library. All structures were confirmed using IR, 1H NMR, 13C NMR and mass spectrometry. The antibacterial potential of twelve selected derivatives was evaluated through molecular docking against the DNA Gyrase B ATP-binding pocket (PDB: 1KZN). Utilizing CB-Dock2, AutoDock Vina, and PLIP analysis, the study predicted binding affinities and interaction profiles. Favourable binding energies and significant intermolecular interactions highlight these thiazine derivatives as promising leads for further antibacterial development. ## Toward AI Virtual Cells for Hepatology: Representation, Generation, Dynamics, and Intervention in Single-Cell Models. - Source: Clinical and molecular hepatology (journals) - Date: 2026-09-17T00:00:00Z - Authors: Youngseok Choi, I. Koh, Murim Choi, Joon-Yong An - Journal: Clinical and molecular hepatology - DOI: 10.3350/cmh.2026.0820 - External ID: 726e90f95d72a4e44ecb3b761da6230dcde0611f - Keywords: Generative models - Source URL: - Dashboard article: Abstract: \`\`Single-cell and spatial atlases describe the healthy and diseased liver at high resolution, including lobular hepatocyte zonation, fibrotic macrophage-stellate niches, cholangiocyte reactions, immune remodeling, and hepatocellular carcinoma ecosystems. These maps show where cell states occur but do not, by themselves, predict whether liver injury will progress or how the liver will respond to an untested drug, toxicant, or genetic perturbation. In this review, we organize current approaches toward an AI Virtual Cell (AIVC) for the liver into three complementary modeling routes. Generative models represent cell states, dynamics and transport models infer state transitions, and pretrained or foundation models test whether learned representations transfer across donors, etiologies, disease stages, and platforms. Perturbation-response prediction serves as a cross-cutting assessment of whether these layers can predict responses to untested genetic, chemical, inflammatory, or metabolic interventions. Available evidence can be categorized as direct liver validation, liver-included benchmarks, general single-cell evidence, and conceptual applications. Published models demonstrate individual components, including atlas integration, inferred trajectories, transferable representations, and retrospective response programs. However, these models do not constitute a prospectively validated liver simulator. At minimum, evaluation should include donor-, etiology-, stage-, platform-, and perturbation-level hold-outs. Model performance should be reported using response direction, recovery of differentially expressed genes and rare states, and calibrated uncertainty. Claims about tissue- or function-level prediction additionally require independent spatial, histologic, metabolic, and functional readouts. Near-term use should prioritize experiment selection and hypothesis generation, whereas clinical decision support remains a longer-term objective. ## Trifluoroacetic Acid Competitively Inhibits Lactate Dehydrogenase to Disrupt Endothelial Glycolysis and Impair Vascular Development - Source: Environmental Science & Technology (journals) - Date: 2026-09-17T00:00:00Z - Categories: Docking & Screening - Authors: Xiao-Lian Cao, Shu-Ting Huang, Chen-Xin Li, Shu-Xin Jiang, Jun-Ru Li, Ying-Ying Zhou, Kai-Qin Huang, Wei-Bang Ma, Yu Chu, Pan-Na Yang, Xiao-Xing Kou, Yi-Tao Pan, Sheng-Tao Ma, Jia-Yin Dai, Xifei Yang, Yan-Hong Wei - Journal: Environmental Science & Technology - DOI: 10.1021/acs.est.6c06569 - External ID: 05183533b4112b6ada158f1ba756f766999426da - Keywords: molecular docking, Enzyme - Source URL: - Dashboard article: Abstract: Trifluoroacetic acid (TFA) is a two-carbon ultrashort-chain per- and polyfluoroalkyl substance (PFAS) and constitutes the most abundant PFAS in the environment. TFA has historically been regarded as a compound of relatively low acute lethality, yet its adverse effects on sensitive targets at environmentally relevant levels are largely unknown. Due to extreme hydrophilicity and anionic form in the aqueous phase of blood, TFA manifests high accessibility to the innermost layer of vasculature. In this study, we investigated the vascular toxicity of TFA using human umbilical vein endothelial cell and zebrafish embryo models at concentrations of 0.8–2000 μg/L. The results showed that TFA exposure reduced ATP content, lactate production, and glycolytic flux, concomitantly increasing pyruvate accumulation. The benchmark dose lower confidence limit (BMDL5) values ranged from 0.44 to 2.65 μg/L, falling within the range of environmental levels. Enzyme kinetics and molecular docking indicated that TFA competitively inhibited lactate dehydrogenase (LDH) at the pyruvate-binding site. In addition, TFA inhibited endothelial migration, induced cellular F-actin disruption, and impaired vascular development in zebrafish larvae. The AI-aided vascular phenomics revealed the most prominent defect in the subintestinal venous plexus. A quantitative adverse outcome pathway was subsequently established, and key-event relationships were robustly fitted, with R2 values exceeding 0.72. This study unravels a metabolic perturbation pathway of TFA-induced vascular toxicity and reinforces the need to reassess the health risks of TFA.