# Chem(o)Info Radar article snapshot > 312 records returned from 3050 current records. Generated: 2026-09-11T22:25:22.065536+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. ## A Machine Learning Framework with Cascade Molecular Representation for Predicting Self-Healing Efficiency in Vitrimers - Source: Digital Discovery (journals) - Date: 2026-09-11T14:23:25Z - Categories: Property Prediction - Authors: Mohammad Hossein Golbabaei, Hunter Garner, Ning Zhang - Journal: Digital Discovery - DOI: 10.1039/d6dd00171h - Keywords: Molecular Representation - Source URL: - Dashboard article: Abstract: Vitrimers are an emerging class of polymeric materials that combine the structural stability of thermosets with the reprocessability and self-healing capabilities of thermoplastics. Despite their promise, accurately predicting self-healing efficiency... ## Molecular docking versus co-folding for protein-peptide pose prediction: effects of protein flexibility and training set overlap - Source: Digital Discovery (journals) - Date: 2026-09-11T13:23:07Z - Categories: Docking & Screening - Authors: Mark Fonteyne, Nada Badr, Peter Kuppen, Alexander L. Vahrmeijer, Gerard Van Westen, Willem Jespers - Journal: Digital Discovery - DOI: 10.1039/d6dd00242k - Keywords: Molecular docking - Source URL: - Dashboard article: Abstract: Synthetic peptides are increasingly important as therapeutic and diagnostic agents due to their high specificity, ease in synthesis, and suitability for targeting protein interfaces. However, accurate prediction of peptide binding... ## moTSart: Accelerating Automated Transition State Search with Generative Models in a Low-Data Regime - Source: Digital Discovery (journals) - Date: 2026-09-11T12:43:07Z - Authors: Leonard Galustian, Johannes Karwounopoulos, Tori Demuth, Jasper De Landsheere, Konstantin Mark, Maximilian Peter-Paul Kovar, Anton Zamyatin, Dennis Svatunek, Esther Heid - Journal: Digital Discovery - DOI: 10.1039/d6dd00259e - Keywords: Generative Models - Source URL: - Dashboard article: Abstract: Transition states (TSs) are first-order saddle points on the potential energy surface, and thus the highest-energy structure on the (multi-step) minimum-energy reaction pathway that corresponds to a chemical reaction, determining... ## Efficient Visualization of Chemical Space - Source: Chemical Science (journals) - Date: 2026-09-11T12:23:19Z - Categories: Cheminformatics - Authors: Akash Surendran, Krisztina Zsigmond, Lexin Chen, Ramon Alain Miranda-Quintana - Journal: Chemical Science - DOI: 10.1039/d6sc03716j - Keywords: cheminformatics, molecular representations - Source URL: - Dashboard article: Abstract: The concept of chemical space is critical in cheminformatics, medicinal chemistry, and machine learning applications. Despite this, the high dimensionality of molecular representations greatly complicates its sampling, analysis, and visualization.... ## A comparative study identifies random forest with minimum redundancy maximum relevance feature selection as a superior transcriptomic classifier for gastric adenocarcinoma diagnosis - Source: Scientific Reports (journals) - Date: 2026-09-11T00:00:00+00:00 - Authors: Zahra Khalili Azni, Matia Sadat Borhani, Hossein Sabouri, Sayed Javad Sajadi, Maryam Pasandideh Arjmand - Journal: Scientific Reports - DOI: 10.1038/s41598-026-67657-w - Keywords: random forest, XGBoost - Source URL: - Dashboard article: Abstract: The high-dimensional nature of transcriptomic data complicates the early diagnosis of gastric adenocarcinoma. While machine learning offers promise, the optimal synergy between feature selection and classifiers is unclear. To define this, we performed a systematic comparison using 1,132 tumor and 131 normal gastric tissue samples from public microarrays. Five feature selection methods (Minimum Redundancy Maximum Relevance or MRMR, F-Test, Chi², Variance Threshold, and random forest importance) and nine classifiers (RF, XGBoost, AdaBoost, SVM, KNN, DT, NB, RUSBoost, NN) were optimized via Bayesian hyperparameter tuning and evaluated using stratified cross-validation and an independent test set. Comprehensive metrics (AUC-ROC, F1, MCC, Accuracy, Precision, Recall, Kappa) identified MRMR as the most effective feature selection method, with detailed comparative results presented. Ensemble classifiers, particularly RF, XGBoost, and AdaBoost, outperformed others. The optimal pipeline combined RF with MRMR feature selection. We conclude that integrating mutual information-based feature selection with ensemble learning yields a high-performance, generalizable transcriptomic classifier, forming a robust foundation for a cost-effective molecular diagnostic tool for early gastric cancer detection. ## A Gentle Push: Fine-Tuning Foundation Models for Complex Oxide Surface Reconstructions - Source: ChemRxiv (preprints) - Date: 2026-09-11T00:00:00Z - Authors: Ralf Wanzenböck, Eva Doloszeski, Georg K. H. Madsen - DOI: 10.26434/chemrxiv.15001629/v3 - External ID: 10.26434/chemrxiv.15001629/v3 - Source URL: - Dashboard article: Abstract: Structure-prediction methods routinely sample highly distorted, far-from-equilibrium configurations. Conventional workflows based on machine learning interatomic potentials (MLIPs) therefore require the inclusion of unphysical local atomic environments in the training data to ensure stability during the search. We show that foundation MLIPs, fine-tuned on minimal datasets containing only physically reasonable structures, can serve as robust backbones for evolutionary structure prediction of complex surface reconstructions. We systematically benchmark fine-tuning strategies and identify that freezing all layers except the readouts as the most effective strategy in the low-data regime. Fine-tuning on as few as 23 structures is sufficient to perform evolutionary searches on reconstructed SrTiO 3 (110) surfaces. For BaTiO 3 (001), we further demonstrate that a lightweight active-learning approach, based on fine-tuning on as few as 20 structures, enables the identification of new low-energy (√5 × √5)R26.6° reconstructions. ## A Nano-QSAR Model for the Cytotoxicity of Doxorubicin Prodrug Micelles in Multidrug-Resistant Cancer Cells - Source: ACS Omega (journals) - Date: 2026-09-11T00:00:00+00:00 - Categories: Property Prediction, ADMET & Safety - Authors: Soniya Kurian, Roy Santiago - Journal: ACS Omega - DOI: 10.1021/acsomega.6c03028 - Keywords: QSAR, pIC50 - Source URL: - Dashboard article: Abstract: Nano-QSAR modeling provides a cost-effective approach for predicting the toxic effects of nanoparticles by reducing the experimental effort and improving the predictive accuracy. Cytotoxicity is the degree of toxicity of the particles toward the cells. In this study, a nano-QSAR model was developed for estimating the cytotoxicity (pIC50) of doxorubicin prodrug micelles on MCF-7/ADR cells by using degree-based topological indices and physicochemical properties such as particle size and polydispersity index. Doxorubicin is a leading anticancer drug, and using it as a nanomedicine, doxorubicin prodrug loaded micelles, improves the therapeutic efficiency, reduces the side effects, and lowers the multidrug-resistance of tumor cells to doxorubicin. Multiple linear regression analysis was used for the model development. The developed model exhibited a meaningful preliminary correlation, with a R2 value exceeding 0.6, between the topological and physicochemical descriptors and the cytotoxicity. As a proof of concept, this proposed model illustrates the potential significance of topological indices as structural descriptors in nano-QSAR modeling for predicting the cytotoxicity of the prodrug micellar system. ## Assessing the performance of implicit ensembling of neural networks for proteochemometric modeling with RandomNets - Source: ChemRxiv (preprints) - Date: 2026-09-11T00:00:00Z - Categories: Property Prediction - Authors: Chiel Jespers, Olivier J. M. Béquignon, Mike Preuss, Gerard J. P. van Westen - DOI: 10.26434/chemrxiv.15008684/v1 - External ID: 10.26434/chemrxiv.15008684/v1 - Keywords: QSAR, ChEMBL, Random Forest - Source URL: - Dashboard article: Abstract: RandomNets, a recently introduced neural-network architecture that emulates the feature and sample bagging of Random Forests through an implicit, vectorised ensemble, has shown promising performance for quantitative structure–activity relationship (QSAR) modelling, but so far only on random data splits. Here we evaluate RandomNets in a proteochemometric (PCM) setting on two benchmarks: an original dataset (BtH) and a newly prepared dataset derived from ChEMBL 35, comprising 571,197 interactions across 382,493 unique compounds and 2,023 unique targets. Models were built from Morgan fingerprints together with additional physicochemical and protein descriptors, and assessed under both random and temporal splits. All models were trained within QSPRpred, enabling reproducible and benchmark-agnostic comparison against the best-performing PCM models from BtH, a deep neural network (DNN\_PCM) and a Random Forest (RF\_PCM). RandomNets with two layers gave the best overall performance, notably improving on the Matthews correlation coefficient (MCC) of the DNN baseline for the temporal split by 5.55% on the original dataset and 8.73% on the new dataset, while taking 2× less time to train on the original dataset and the same time on the new dataset. These results affirm the potential of RandomNets and extend it to PCM modelling. ## Automatic Text-fishing, Unification, and Normalization Application in Python (aTUNApy): a tool to build molecular databases - Source: ChemRxiv (preprints) - Date: 2026-09-11T00:00:00Z - Categories: Cheminformatics, LLMs & Agents - Authors: Aylin Del Moral Morales, José L. Medina-Franco - DOI: 10.26434/chemrxiv.15008687/v1 - External ID: 10.26434/chemrxiv.15008687/v1 - Keywords: SMILES, RDKit, PubChem, GPT, LLM, Claude, bioactivity - Source URL: - Dashboard article: Abstract: The construction and maintenance of chemical databases from the scientific literature remain labor-intensive, largely because relevant information is dispersed across thousands of publications and compounds are reported under inconsistent nomenclature. Here we present aTUNApy (automatic Text-fishing, Unification, and Normalization Application in Python), an open-source tool that couples large language model (LLM)-based extraction of compound information from PubMed abstracts with nomenclature harmonization through PubChem and SMILES standardization with RDKit. aTUNApy retrieves articles for a user-defined query, extracts compound names together with user-specified attributes such as bioactivity, mechanism or source, and maps each mention to a unique PubChem identifier, collapsing synonyms onto a single canonical entry. Four LLM families are supported: GPT, Claude, LLaMA and Gemini. We benchmarked all four against a manually curated reference set of 507 articles, including 100 negative controls, covering 148 unique compounds. Claude achieved the highest performance (precision 0.98, recall 0.95, F1 0.97), while Gemini reached comparable precision (0.98) and the lowest hallucination rate (0.2%) at no API cost, reducing a five-day manual curation task to minutes or hours of automated processing. Finally, to demonstrate end-to-end utility, we applied aTUNApy to 8,273 articles from PubMed and Web of Science to build NutriEpiDB, a database of 653 dietary compounds with reported epigenetic effects across 917 food sources. ## Beyond the Hydrophobic Trough of Pf AMA1: Pivotal Roles of Cryptic Loop Dynamics Inform Anti-Malarial Therapeutic Design - Source: Journal of Medicinal Chemistry (journals) - Date: 2026-09-11T00:00:00+00:00 - Authors: Suman Sinha - Journal: Journal of Medicinal Chemistry - DOI: 10.1021/acs.jmedchem.6c00933 - Source URL: - Dashboard article: Abstract: Protein loops often mediate dynamic, conformation-dependent molecular recognition. This perspective examines the essential interaction between Plasmodium falciparum apical membrane antigen 1 (PfAMA1) and rhoptry neck protein 2 (PfRON2), which drives moving-junction formation and erythrocyte invasion. We focus on the PfAMA1 domain 2 (D2) loop and its role in regulating native and therapeutic peptide binding. D2-loop closure creates a cryptic subpocket, termed Hotspot-3, through hydrophobic packing reinforced by hydrogen bonds. This conformational heterogeneity complicates structure-based inhibitor design because binders must either stabilize preferred loop states or accommodate loop dynamics. These findings highlight residence-time optimization, rather than affinity alone, as a key objective for next-generation PfAMA1 inhibitors. Small molecules may also target PfAMA1, although the ligandability of Hotspot-3 remains unproven. Strategies combining N-terminal helicity, cryptic-pocket engagement, conformational-ensemble modelling, and machine learning may enable longer-acting and potentially strain-transcending antimalarial therapeutics. However, Hotspot-3 conservation and sustained inhibition require validation across divergent PfAMA1 alleles. ## Chemically Informed Machine Learning Approach for Prediction of Reactivity Ratios in Radical Copolymerization - Source: ACS Omega (journals) - Date: 2026-09-11T00:00:00+00:00 - Authors: Habibollah Safari, Mona Bavarian - Journal: ACS Omega - DOI: 10.1021/acsomega.6c00795 - Source URL: - Dashboard article: Abstract: Predicting monomer reactivity ratios is crucial for controlling monomer sequence distribution in copolymers and their properties. Traditional experimental methods of determining reactivity ratios are time-consuming and resource-intensive, while existing computational methods often struggle with accuracy or scalability. Here, we present a method that combines unsupervised learning with artificial neural networks to predict reactivity ratios in radical copolymerization. By applying spectral clustering to physicochemical features of monomers, we identified three distinct monomer groups with characteristic reactivity patterns. This computationally efficient clustering approach revealed specific monomer group interactions leading to different sequence arrangements including alternating, random, block, and gradient copolymers─providing chemical insights for initial exploration. Building upon these insights, we trained artificial neural networks to achieve quantitative reactivity ratio predictions. We explored two integration strategies: direct feature concatenation and cluster-specific training, which demonstrated performance enhancements for targeted chemical domains compared to general training with equivalent sample sizes. However, models utilizing complete data sets outperformed specialized models trained on focused subsets, revealing a fundamental trade-off between chemical specificity and data availability. This work demonstrates that unsupervised learning offers rapid chemical insight for exploratory analysis, while supervised learning provides the accuracy necessary for final design predictions, with optimal strategies depending on data availability and application requirements. ## Docking-score landscapes shape active-learning performance across Vina, Glide, and SILCS - Source: Journal of Computer-Aided Molecular Design (journals) - Date: 2026-09-11T00:00:00+00:00 - Authors: Joseph Chung, Aashish Bhatt, Jacob Ede Levine, Yi-Chun Lin, Mingtian Zhao, Alexander D. MacKerell, Sunhwan Jo, Sai Chandra Kosaraju, Yun Lyna Luo - Journal: Journal of Computer-Aided Molecular Design - DOI: 10.1007/s10822-026-00927-x - Keywords: virtual screening, chemical diversity - Source URL: - Dashboard article: Abstract: The rapid expansion of large chemical libraries has created a need for virtual screening workflows that are both efficient and accurate. Active learning (AL) offers a scalable strategy by iteratively training surrogate models to prioritize promising compounds and reduce the number of required docking calculations. However, direct benchmarking of active-learning protocols across docking engines remains limited. In this study, we ask whether docking scores generated by different docking engines affect AL performance and what factors underlie these differences. To do so, we compare four active-learning virtual screening workflows, Vina-MolPAL, Glide-MolPAL, SILCS-MolPAL, and Schrödinger’s active-learning Glide, across multiple protein targets and library sizes. Performance was assessed by recovery of top-scoring molecules according to each workflow’s respective docking engine, score-prediction accuracy, chemical diversity, and computational cost. Vina-MolPAL achieved the highest top-1% recovery at a 1% batch size, whereas SILCS-MolPAL achieved comparable recovery at a larger batch size. Latent-embedding analyses suggest that the docking-score landscape strongly influences active-learning performance. In addition, combining active learning with SILCS provides a computationally efficient, membrane-aware approach for screening compounds at transmembrane binding sites. ## Fourier-based chemometric resolution of the FOLFOX regimen through digital filtering with poly-metric sustainability assessment - Source: Scientific Reports (journals) - Date: 2026-09-11T00:00:00+00:00 - Authors: Hadir M. Maher, Salma Mahmoud Mohamed, Ekram M. Hassan, Amira Fawzy El-Yazbi - Journal: Scientific Reports - DOI: 10.1038/s41598-026-67979-9 - Keywords: chemometric - Source URL: - Dashboard article: Abstract: The FOLFOX regimen, comprising oxaliplatin (OXA), 5-fluorouracil (5-FU), and leucovorin (LU), is a standard first-line protocol for colorectal cancer therapy. Given its clinical centrality, developing rapid and high-throughput analytical tools for simultaneous determination of its components is essential. However, severe spectral overlap among the three drugs severely complicates their direct spectrophotometric analysis. In this study, novel spectrophotometric methods were developed and validated for the concurrent quantification of OXA, 5-FU, and LU without preliminary separation steps. While direct and first-derivative spectrophotometric analysis successfully resolved binary fractions, complete resolution of the ternary mixture was uniquely achieved through a hybrid double divisor ratio spectra (HDDR) approach integrating trigonometric Fourier functions. The methods were validated in accordance with ICH Q2(R1) guidelines, exhibiting high linearity (r > 0.999), excellent recoveries (98.0–102.0%), and high precision (RSD < 2.0%) across therapeutic concentration ranges. Furthermore, multi-metric evaluation tools demonstrated improved scores relative to reported chromatographic methods, including higher AES, AGREE, MoGAPI, AGSA, CaFRI, RGB, BAGI, and VIGI scores. The developed HDDR method offers an eco-friendly, rapid, and cost-effective alternative for routine quality control of the FOLFOX regimen. ## Investigation of the removal of indigo carmine from aqueous solution using ethylenediaminetetraacetic acid-functionalized magnetic 3-aminopropyl triethoxysilane nanocomposites - Source: Reaction Kinetics, Mechanisms and Catalysis (journals) - Date: 2026-09-11T00:00:00+00:00 - Categories: Cheminformatics - Authors: Aurangzeb Junejo, Irfan Ahmed Abbasi, Du Ri Park, Siraj Ahmed, Ick Tae Yeom - Journal: Reaction Kinetics, Mechanisms and Catalysis - DOI: 10.1007/s11144-026-03239-1 - Keywords: indigo - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Learning Memory and Transferability in Coarse-Grained Dynamics - Source: Journal of Chemical Theory and Computation (journals) - Date: 2026-09-11T00:00:00+00:00 - Authors: Shuyuan Zhang, Baocai Jing, Ting Ye - Journal: Journal of Chemical Theory and Computation - DOI: 10.1021/acs.jctc.6c01063 - Keywords: force field - Source URL: - Dashboard article: Abstract: We develop a transferable non-Markovian intelligent dissipative particle dynamics (TNM-IDPD) model that resolves two critical challenges in coarse-grained modeling: the recovery of system dynamics and transferability across thermodynamic states. The model uses deep neural networks to directly learn the mean force field and memory kernel from atomistic simulation data, capturing both static structure and non-Markovian dynamics. It employs Bayesian active learning with minimal human annotation to achieve efficient transferability across thermodynamic states, where predictive uncertainty strategically guides data acquisition by selecting the most informative configurations for manual labeling and identifying reliable predictions for pseudolabeling. Applied to a star polymer system, TNM-IDPD accurately captures static and dynamic properties across varying temperatures and densities, establishing a high-fidelity and data-efficient framework that unifies dynamic accuracy with state transferability. ## Machine Learning Assisted Raman Spectroscopic Identification of Edible Oils in Fried Food Matrix - Source: ChemRxiv (preprints) - Date: 2026-09-11T00:00:00Z - Categories: Spectra & Analytical - Authors: Amrita Shaw, Deepak L. N. Kallepalli, Chandrasekar S. N., Sai Muthukumar V., Jhinuk Gupta - DOI: 10.26434/chemrxiv.15008698/v1 - External ID: 10.26434/chemrxiv.15008698/v1 - Keywords: Random Forest, XGBoost - Source URL: - Dashboard article: Abstract: Identification of the type of cooking oil used in ready-to-eat (RTE) foods is paramount, as it directly impacts consumers’ health. Comprehensive evaluation of existing protocols reveals a pressing need for developing rapid methodologies with improved operational efficiency. This proof-of-concept study reports development of one such rapid, solvent-free analytical framework by coupling Raman spectroscopy with machine learning. It demonstrates the feasibility of oil type identification from fried food matrices without solvent extraction. Five edible oils (sunflower, soybean, groundnut, palm, and vanaspati) and one complex food matrix (potato chips) were chosen for the study. Five types of paper based absorbents were screened for most effective isolation of the oil from the potato chips. Seven supervised classification algorithms, Logistic Regression, SVC (RBF/Linear), LightGBM, Random Forest Classifier, XGBoost, and KNN were evaluated using stratified five-fold cross-validation. Model performance was assessed using four key metrics: accuracy, precision, recall, and F1-score. Coarse hyperparameter tuning was employed to further enhance the model performance. Comparative analysis of the tested algorithms revealed SVC (RBF) to achieve the best prediction accuracy (97% for oils and 89.8% for chips). The framework reduces the total analysis time from hours to minutes by replacing conventional oil extraction with the newly optimized sampling technique. This new sampling strategy was validated using cross-correlation analysis and non-negative least squares (NNLS) spectral decomposition. The high correlation coefficient (ρ= 0.91) and high spectral contribution (56%) of oils in the chips spectra confirmed the preservation of the oil fingerprints within the chips spectra. ## Machine learning-accelerated analysis of in utero embryo phenotyping in C. elegans for reproductive toxicity assessment - Source: Scientific Reports (journals) - Date: 2026-09-11T00:00:00+00:00 - Authors: Abhishri Medewar, Andrew DuPlissis, Adam Laing, Amber Shen, Evan Hegarty, Sebastian Gomez, Gina Carrion, Julia Brown, Sudip Mondal, Adela Ben-Yakar - Journal: Scientific Reports - DOI: 10.1038/s41598-026-70459-9 - Keywords: autoencoder - Source URL: - Dashboard article: Abstract: Predictive new approach methodologies (NAMs) for developmental and reproductive toxicity (DART) assessment are increasingly needed as reliance on conventional mammalian studies decreases and chemical safety evaluation demands continue to expand. Whole-organism NAMs, including Caenorhabditis elegans , provide a scalable non-mammalian strategy because they preserve conserved biological pathways within an intact physiological system. We recently developed vivoDART, a rapid and repeatable C. elegans assay that quantifies in utero embryo development and overcomes key limitations of traditional labor-intensive, multiday C. elegans DART workflows. However, despite its robustness and reproducibility, vivoDART still requires manual analysis of tens of thousands of embryos per chemical, a process that is time-consuming and prone to user-dependent variability. To address this bottleneck, we developed EmbryoMAE-Det, a machine-learning framework trained on ~ 48,000 manually segmented embryos from 1,547 worms. The model combines self-supervised masked autoencoder pretraining with supervised object detection and classification to identify embryos within the C. elegans uterus and classify them by developmental stage. EmbryoMAE-Det achieved high accuracy (mAP = 88.7%), with AP values of 92.8% and 84.7% for early- and late-stage embryo counts, respectively. Model-derived embryo counts showed low variability with CV%s for technical replicates below 10.4%, sufficient statistical power to detect changes as small as 5–18%, and EC 50 values statistically indistinguishable from those obtained by manual scoring. The fully automated workflow reduces analysis time by 1,000× to ~ 20 min per chip, which is shorter than the data acquisition time, and thus removes the analysis bottleneck in the assay. In summary, this work establishes an integrated whole-organism imaging and machine-learning platform for rapid, reproducible, and high-content DART evaluation using C. elegans as a NAM. ## Physics-Inspired Multi-Body Descriptors Enable Interpretable Density Prediction of Energetic Ionic Salts - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-11T00:00:00+00:00 - Categories: Cheminformatics, Property Prediction - Authors: Xiaoyan Wang, Xiaokai He, Yining Zhang, Guojin Li, Chao Chen, Yingzhe Liu - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c02052 - Keywords: molecular descriptors - Source URL: - Dashboard article: Abstract: Machine learning has introduced a transformative paradigm for the rapid prediction of crystal properties, circumventing the need for explicit knowledge of crystal packing structures. However, when applied to high-density organic systems such as energetic ionic salts, conventional general-purpose molecular descriptors often exhibit limited prediction accuracy, generalization capability, and extrapolation capability. To address this challenge, we propose a physics-inspired multi-body descriptor (PIMBD). By decoupling the modeling of cations and anions and employing the co-evolutionary optimization of multi-body function hyperparameters, PIMBD effectively captures spatial arrangement features of atoms governed by strong Coulombic interactions. We established a benchmark dataset comprising more than 1200 energetic ionic crystals and conducted a systematic comparison of 6 representative descriptor frameworks and 29 machine learning algorithms. The results show that PIMBD achieves high predictive accuracy, with a mean absolute error (MAE) of 0.034 g/cm3. Meanwhile, it achieves an MAE of 0.0282 g/cm3 on an independent validation set composed of pentazolate-based ionic salts, shows pronounced advantages in the high-density region, and outperforms other descriptors in terms of generalization robustness and extrapolation capability. Furthermore, atomic-scale interpretability analyses were performed to elucidate the critical structural factors governing crystal density, leading to the distillation of three physically meaningful design principles. This work provides an efficient and interpretable descriptive methodology specifically tailored for ionic salts and may potentially accelerate the rational structure design of high-energy-density materials. ## UniMolRep: A Python Package for AI-Oriented Molecular Representation Modeling - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-11T00:00:00+00:00 - Categories: Cheminformatics, Property Prediction - Authors: Weiqing Guo, Jiawei Chen, Debby D. Wang - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c01681 - Keywords: Molecular representation, molecular representations - Source URL: - Dashboard article: - Code: Abstract: Summary: Molecular representation modeling is a crucial component in computational biology. However, existing tools often have limited coverage of molecular representation models. UniMolRep is a comprehensive Python toolkit for generating molecular representations in multiple dimensions for a variety of machine learning prediction tasks. In addition, it provides an efficient and user-friendly interface. Availability and implementation:UniMolRep is available as a GitHub repository https://github.com/wguo0209/MolRep\_Toolkit. Detailed API references and full documentation are provided within the repository. Contact: For further information, please contact at wguo@hkmu.edu.hk. ## From Document Silos to Process Intelligence: A Multi-Layer Knowledge Graph for CMC Process Development - Source: arXiv (preprints) - Date: 2026-09-10T12:59:04Z - Categories: LLMs & Agents - Authors: Reza Amirmoshiri, Faryad Sahneh, Yasser Jangjou - External ID: 2609.11493v1 - Keywords: LLM - Source URL: - Dashboard article: - PDF: Abstract: Chemistry, Manufacturing and Controls (CMC) process development generates an enormous body of technical information across a multi-stage, knowledge-intensive continuum from drug discovery to commercial manufacturing. This knowledge is traditionally fragmented across functions and heterogeneous formats, causing traceability gaps and significant knowledge-management costs during technology transfer and regulatory filing. We present a modular agentic-AI platform that converts a heterogeneous corpus of process-development documents into a queryable, dual-layer knowledge graph. A base knowledge layer builds a lexical graph with a Document-Section-Chunk hierarchy through lossless ingestion of digital, scanned, handwritten, and multilingual documents, while an intelligence layer extracts ontology-aligned entities and bridges cross-document concepts through a provenance-anchored domain graph. LLM agents operate across both layers, selecting the retrieval path best suited to each question. We evaluate the lexical layer with a novel three-tier protocol measuring the deployment-fidelity of a retrieval-augmented generation (RAG) system on proprietary data, demonstrated on 505 questions curated from 38 development reports of a Sanofi small-molecule program. Tier-1 multiple-choice accuracy of 95% signals strong platform reliability; the stricter Tier-2 LLM-judge pass rate of 85%, which degrades on comparative and corpus-wide questions, reveals a failure taxonomy that Tier-1 accuracy alone fails to capture. A router agent selects between layers according to question type. We anticipate this protocol will enable future designers of agentic platforms to assess their systems against nonpublic databases, and that graph-based architectures will see broader adoption in pharma as a means of transforming fragmented document repositories into structured process intelligence. ## Coherent Floquet quantum reservoirs for molecular property prediction - Source: arXiv (preprints) - Date: 2026-09-10T04:27:01Z - Categories: Property Prediction, ADMET & Safety - Authors: Luofei Wang, Da Zhang, Congren Wang, Yiming Li, Yuxiao Yang, Xuan Zhang, Xuefeng Cui, Zhang-Qi Yin - External ID: 2609.11071v1 - Keywords: property prediction, molecular property - Source URL: - Dashboard article: - PDF: Abstract: Quantum reservoir computing (QRC) uses quantum dynamics to represent input histories for prediction through a trained classical readout. Discrete time crystals (DTCs) exhibit robust subharmonic responses under periodic driving, and previous work has used their dynamics to construct DTC-QRC. Here we construct a DTC-based reservoir architecture to predict molecular properties from structural and dynamical observations. Coherent Floquet evolution processes local molecular graph events and surface-hopping frames, while controlled reset regulates the contribution of earlier inputs. Measurements at the end of each input sequence yield a feature vector of fixed dimension. Trained classical decoders use this vector for inhibitor-activity and blood--brain-barrier permeability classification and electronic-gap forecasting, while the reservoir parameters remain fixed during training. With matched input lengths and output widths, DTC-QRC outperforms echo-state networks on long-prefix graph classification and the studied ethene gap forecasting tasks. Dephasing lowers performance in both applications, consistent with a role for coherent propagation. Experiments on the Quafu superconducting quantum cloud platform show that pair observables retain task information under device noise. The architecture provides a common framework for molecular screening and time-resolved property prediction using quantum reservoir computing. ## A Task-Adaptive Multimodal Pretrained Framework for Antibiotic Virtual Screening with Joint Activity and Cytotoxicity Prediction - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-10T00:00:00+00:00 - Categories: Cheminformatics, Property Prediction, Docking & Screening, ADMET & Safety - Authors: Zhijie Pan, Xin Yang, Wenchi Ge, Churong Wang, Jianqiang Sun, Qi Zhao - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c02659 - Keywords: molecular fingerprints, Virtual Screening, activity prediction, molecular representations - Source URL: - Dashboard article: Abstract: Antimicrobial resistance (AMR) poses a growing threat to global public health, particularly for priority pathogens such as Staphylococcus aureus (SA) and Neisseria gonorrheae (NG), for which therapeutic options are increasingly limited. Although deep learning has accelerated antibiotic discovery, existing approaches are often constrained by single-modality molecular representations and a lack of joint modeling for antibacterial activity and host cell toxicity. Here, we present MAPViS, a task-adaptive multimodal pretrained framework for jointly predicting antibacterial activity against SA and NG together with cytotoxicity risk in three human cell lines (HepG2, HSkMC, and IMR-90). MAPViS integrates molecular graph representations, molecular fingerprints, and descriptor/3D geometric features through a gated fusion mechanism and combines large-scale self-supervised pretraining with supervised task-specific fine-tuning. The graph encoder is pretrained on approximately 10 million molecules using hierarchical pseudolabel classification and masked graph contrastive learning to improve representation quality and transferability. On benchmark datasets derived from experimentally screened compounds, MAPViS achieves competitive performance across antibacterial activity prediction tasks and shows advantages in early recognition metrics relevant to virtual screening, including AUPRC, Precision@30, and EF@1.0%. External validation further shows strong top-ranked enrichment on the SA task, where all Top-10 candidates are experimentally active, whereas performance on NG is more modest, indicating task-dependent generalization. Overall, MAPViS provides a scalable multimodal framework for joint activity–toxicity modeling and offers a practical computational tool for antibiotic virtual screening and candidate prioritization. ## Accelerating Proton Affinity Prediction with Multi-Fidelity Machine Learning - Source: ChemRxiv (preprints) - Date: 2026-09-10T00:00:00Z - Categories: Cheminformatics - Authors: Debjyoti Bhattacharya, Yifan Liu, Valentino R. Cooper, Wesley F. Reinhart - DOI: 10.26434/chemrxiv.15002556/v2 - External ID: 10.26434/chemrxiv.15002556/v2 - Keywords: cheminformatics, Claude, DFT, B3LYP - Source URL: - Dashboard article: Abstract: Accurate gas-phase proton affinities (PAs) are essential for designing proton carriers in anhydrous fuel cells, yet high-level quantum-chemical calculations are too expensive for large libraries. We develop a multi-fidelity Δ-learning framework that corrects PM7 semi-empirical estimates toward high-fidelity references using only PM7 electronic properties and cheminformatics descriptors, without DFT-derived features. On a 1,155-molecule NIST WebBook benchmark, an ExtraTrees model correction reduces the PM7 mean absolute error (MAE) from 8.21 to 2.87 ± 0.26 kcal/mol, a 47% reduction over a calibrated PM7 baseline of 5.44 ± 0.39 kcal/mol. A DFT-descriptor-based model trained to the same experimental targets yields 3.17 ± 0.24 kcal/mol, demonstrating that PM7-based features achieve comparable accuracy while avoiding DFT cost at inference. We extend Δ-learning to the harder task of site-specific prediction against a B3LYP/def2-TZVP reference, across 251 diverse molecules and 821 site-matched protonation sites, reducing the raw PM7 error from 10.26 to 7.44 ± 1.00 kcal/mol, versus 8.23 ± 1.02 kcal/mol for calibrated PM7. Functional-group and SHAP analyses identify correction patterns and high-variance cases where direct DFT remains preferable. We then apply the RandomForest model in a seven-stage workflow screening for candidate anhydrous proton carriers, combining similarity retrieval, PM7+Δ-ML prediction, Claude Opus 4.6-assisted preliminary donor–acceptor motif screening, and Pareto selection. From 821,435 filtered ZINC molecules, 30 candidates were DFT-validated, of which 29 fall within the target window (210–235 kcal/mol). Applied prospectively to these 30 molecules, the correction lowers the raw PM7 error from 13.51 to 3.99 kcal/mol. The DFT PA window, motif check, and functional-group exclusion retain 26 candidates, and the five ranked highest by DFT proton affinity span 226.6–234.3 kcal/mol, at or above the 225.3 kcal/mol PA of imidazole. These single-molecule calculations prioritize candidates for experiment but do not by themselves establish condensed-phase proton conductivity. ## Chiral gold nanofibers as substrates towards SERS-based chiral and flexible sensing - Source: ChemRxiv (preprints) - Date: 2026-09-10T00:00:00Z - Categories: Cheminformatics - Authors: Mathias R. S. Nielsen, Christian Hirsch, Gohar Soufi, Anja Boisen, Miguel Alexandre Ramos-Docampo - DOI: 10.26434/chemrxiv.15008635/v1 - External ID: 10.26434/chemrxiv.15008635/v1 - Keywords: molecular fingerprints - Source URL: - Dashboard article: Abstract: Chiral plasmonic nanostructures offer unique opportunities for combining surfaceenhanced Raman scattering (SERS) with enantioselective molecular recognition. However, their integration into practical sensing platforms remains challenging. Here, we demonstrate high-aspect-ratio chiral gold nanofibers as versatile substrates for sensitive, chiral, and flexible SERS sensing. Using trans-1,2-bis(4-pyridyl)ethene as a model analyte, the chiral gold nanofibers enable molecular detection down to 1 nM and exhibit higher SERS efficiency (between 2-3× higher) than benchmark nanomaterials such as chiral and achiral gold nanorods. Beyond molecular detection, the chiral gold nanofibers generate distinct SERS responses towards D- and L-cysteine and mixtures with different enantiomeric compositions above analyte concentrations of, at least, 10 mM. Finally, the chiral gold nanofibers retain their SERS activity after integration into low-cost, flexible paper scaffolds, enabling spatial mapping and recovery of characteristic molecular fingerprints across the heterogeneous surface. These results establish chiral gold nanofibers as potential multifunctional plasmonic building blocks that combine molecular sensitivity, enantioselective recognition, and compatibility with flexible surfaces, providing a route towards adaptable platforms for label-free chiral sensing. ## Closed-Loop Multi-Objective Optimization for Receptor-Selective Cell-Penetrating Peptide Design - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-10T00:00:00+00:00 - Categories: Design de novo - Authors: Iori Yamahata, Hidetomo Yokoo, Yosuke Demizu, Teppei Shimamura, Shuto Hayashi - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c01267 - Keywords: multi objective optimization, generative model, Receptor, molecular dynamics - Source URL: - Dashboard article: Abstract: Cell-penetrating peptides (CPPs) can deliver diverse cargoes into cells. However, designing CPPs with receptor-selective interaction profiles remains difficult because interactions with individual cell-surface components cannot be tuned independently. Here, we developed a closed-loop in silico framework for receptor-selective CPP design, in which receptor interactions are formulated as explicit objectives in a multi-objective optimization problem. We first constructed a CPP-like candidate library by using a sequence generative model fine-tuned on known CPPs. The framework then evaluated candidate peptides by receptor-wise docking, molecular dynamics simulations, and MM/GBSA to compute receptor-wise binding scores as computational proxies for receptor-associated interactions. These scores were used iteratively to propose subsequent candidates by multi-objective Bayesian optimization. Applied to a CXCR4/NRP1 design setting, the framework identified candidates with more favorable predicted interaction profiles characterized by higher CXCR4 binding scores and lower NRP1 binding scores. We selected 10 peptides from the computationally identified candidates for cell-based imaging and found that 4 showed higher enrichment in CXCR4-positive regions than in NRP1-positive regions under the tested conditions. These results show that the proposed framework provides a practical in silico approach for designing CPPs with receptor-selective interaction profiles. ## Comparative Adsorption of Mono-, Di-, and Tricarboxylic Acids on Magnetically Modified Graphene Oxide: Influence of Molecular Structure on Adsorption Behavior - Source: ACS Omega (journals) - Date: 2026-09-10T00:00:00+00:00 - Authors: Ebubekir Ekinci, Hasan Uslu, Şahika Sena Bayazit, Mehmet Yetişen - Journal: ACS Omega - DOI: 10.1021/acsomega.6c05759 - Keywords: RL, chemicals - Source URL: - Dashboard article: Abstract: This study investigates the efficient adsorption and recovery of biotechnologically significant organic acids (lactic, succinic, and citric acids) using a magnetically modified graphene oxide (MGO) composite. The MGO was synthesized via chemical coprecipitation, and its structural, functional, and textural properties were comprehensively characterized using X-ray diffraction (XRD), Fourier transform infrared spectroscopy (FTIR), and Brunauer–Emmett–Teller (BET) analysis. Nitrogen adsorption–desorption measurements revealed a mesoporous structure with a specific surface area of 37.94 m2/g, a total pore volume of 0.078 cm3/g, and an average pore radius of 4.12 nm, which facilitate effective molecular diffusion into the internal pore framework. Batch adsorption experiments were conducted at 25 °C using an adsorbent dosage of 10 mg in a 20 mL solution volume across a concentration range of 10–100 g/L to evaluate equilibrium performance in comparison with a magnetically modified activated carbon (AC) reference. The equilibrium times were established as 50–80 min depending on the specific organic acid. Adsorption kinetics were accurately described by the pseudo-second-order (PSO) model (R2 > 0.99), confirming a chemisorption-controlled mechanism for all investigated acids. To ensure physical consistency and rigorous mass balance reconciliation, adsorption capacities were evaluated using mass-normalized (mg/g) units. The equilibrium data were analyzed using Langmuir, Freundlich, and Temkin isotherm models. While the results showed excellent agreement with the Langmuir model, the calculated dimensionless separation factor (RL) values (ranging from 2.3 × 10–6 to 2.8 × 10–3) further confirmed that the adsorption process is highly favorable for all investigated acids across the concentration range. Lactic acid exhibited the highest maximum adsorption capacity (qmax = 334.45 mg/g), which is attributed to its smaller molecular size and the specific interaction of its hydroxyl (−OH) group, enabling enhanced hydrogen bonding and electrostatic interactions with the oxygen-rich surface sites of the MGO. Furthermore, regeneration studies demonstrated that the MGO composite maintains a high capacity retention rate of 91.4% after five consecutive cycles, confirming its excellent structural stability and long-term industrial feasibility. The results demonstrate that the MGO composite provides superior adsorption performance compared to AC and allows for rapid magnetic separation from aqueous solutions. This study underscores the potential of functionalized graphene-based materials for the sustainable separation and purification of platform chemicals in the biotechnology industry. ## Comparative Evaluation of Graph Neural Networks for Molecular Lipophilicity Prediction - Source: ChemRxiv (preprints) - Date: 2026-09-10T00:00:00Z - Categories: Cheminformatics - Authors: Annika Bande, Sandeep Kumar - DOI: 10.26434/chemrxiv.15008627/v1 - External ID: 10.26434/chemrxiv.15008627/v1 - Keywords: cheminformatics, Graph Neural Networks, GNNs, GNN, molecular representations - Source URL: - Dashboard article: Abstract: Predicting molecular properties is a central challenge in cheminformatics and plays a critical role in drug discovery and materials science. Lipophilicity, which describes the partitioning of a molecule between lipid and aqueous environments, is a key physicochemical property influencing drug behavior. Although quantum mechanical methods can provide accurate predictions, they are often computationally demanding. In contrast, large experimental databases can be leveraged to train data-driven models such as graph neural networks (GNNs), which learn molecular representations directly from chemical structures. In this study, we compare four GNN architectures—graph convolutional networks (GCN), graph isomorphism networks (GIN), Graph Sample and AggregatE (GraphSAGE), and Attentive Fingerprint (Attentive FP)—for predicting lipophilicity using both regression and classification approaches. The models are evaluated on a benchmark dataset under consistent preprocessing, optimization, and training conditions. Our results show that the Attentive FP model consistently outperforms the other architectures, yielding improved predictive accuracy across both tasks. We further examine the impact of outlier removal during preprocessing, observing reduced prediction errors and improved robustness. These findings provide insight into the capabilities of representative GNN models for molecular lipophilicity prediction. ## DEEP LEARNING HELMHOLTZ ENERGIES FOR THERMODYNAMIC PROPERTY PREDICTION - Source: ChemRxiv (preprints) - Date: 2026-09-10T00:00:00Z - Authors: Maximilian Fleck, Marcelle B M Spera - DOI: 10.26434/chemrxiv.15008626/v1 - External ID: 10.26434/chemrxiv.15008626/v1 - Keywords: PROPERTY PREDICTION - Source URL: - Dashboard article: Abstract: Based on SAFT equations of state (EoS), we propose a physics informed neural network that learns the residual Helmholtz free energy, yielding pressure, chemical potentials, entropy, and phase equilibria by automatic differentiation. Transport properties are connected through additional relations. The model can be pretrained on synthetic EoS data and fine-tuned with heterogeneous experimental data. The model operates on logarithmic scale and is designed to stabilize learning properties that span several orders of magnitude between phases, state points, and molecules. A differentiable transport model further allows viscosity and self-diffusion data to learn the temperature derivative of the Helmholtz energy during fine-tuning. As a proof of concept, we fine-tune the model to experimental temperature–pressure–density, viscosity, and self-diffusion data for water. Despite using only state point information and transport data for this single, particularly challenging fluid, the model predicts vapor pressure, vaporization enthalpy, density and entropy anomalies, and the critical temperature without direct exposure to these properties during fine-tuning. This demonstrates that jointly learning Helmholtz energies and transport coefficients is viable and beneficial. We argue that such Helmholtz energy centered, data driven deep learning models provide an alternative to the current proliferation of isolated, single property machine learning solutions. Generalization to unseen molecules and a fixed functional mode with optimized molecule representation parameters are also demonstrated. ## DeepLabCut-based automated system reveals diverse temperature tolerance among medaka strains and related Oryzias species - Source: Scientific Reports (journals) - Date: 2026-09-10T00:00:00+00:00 - Authors: Yoshiya Matsuo, Takuya Kato, Kiyoshi Naruse, Takashi Yoshimura, Tatsuhito Hasegawa, Tomoya Nakayama - Journal: Scientific Reports - DOI: 10.1038/s41598-026-66712-w - Keywords: classification model - Source URL: - Dashboard article: Abstract: Temperature is a critical environmental factor influencing the physiology and behavior of ectothermic animals, yet conventional methods for evaluating thermal tolerance in fish rely on subjective manual observation of loss of equilibrium (LOE), limiting experimental throughput and introducing observer bias. Here, we developed an automated temperature tolerance evaluation system integrating DeepLabCut-based pose estimation with custom image processing algorithms to objectively quantify the timing of LOE during thermal stress tests. Our system incorporated region partitioning and color transformation preprocessing to improve keypoint detection accuracy, followed by a classification model combining ResNet34-based frame features with keypoint coordinates to objectively determine the timing of LOE without manual observation. Validation against manual annotation showed that the automated system achieved an accuracy comparable to the natural variability between trained investigators, and outperformed naive human observers, supporting its validity as an objective and reproducible alternative to manual scoring. Using this system, we characterized cold and heat tolerance across six medaka strains ( Oryzias latipes : d-rR/TOKYO, HB11A, OK-Cab, HO5 and HdrR-II1; O. sakaizumii : HNI-II). Cold and heat tolerance assessment revealed inter-strain variation, with HdrR-II1 among the most cold- and heat-tolerant strains and HNI-II the least tolerant of both cold and heat stress. We further evaluated cold tolerance in medaka-related species ( O. sinensis , O. cabaranensis , O. curvinotus , O. luzonensis , O. celebensis , and O. javanicus ) and zebrafish ( Danio rerio ), revealing substantial interspecific variation that broadly corresponded with latitudinal distribution. O. latipes , distributed at the highest latitudes among the tested species, exhibited the greatest cold tolerance, whereas O. celebensis , O. javanicus , and other tropical or low-latitude species showed comparatively low cold tolerance. Our automated system provides a robust, high-throughput platform for thermal tolerance evaluation and, combined with the genetic and genomic resources available in medaka, establishes a foundation for elucidating the molecular mechanisms underlying temperature adaptation in fish. ## Domain-specific dataset enable accurate MLIPs for disordered halide-based solid electrolytes - Source: Machine Learning: Science and Technology (journals) - Date: 2026-09-10T00:00:00+00:00 - Authors: Chiku Parida, Arghya Bhowmik, Juan Maria García Lastra - Journal: Machine Learning: Science and Technology - DOI: 10.1088/2632-2153/aea5d8 - Keywords: MLIP, MACE, density functional theory, DFT - Source URL: - Dashboard article: Abstract: Efficient exploration of chemical space with machine-learning interatomic potentials (MLIPs) requires a comprehensive, application-focused training database. Solid-state electrolytes (SSEs) for Li-ion batteries have attracted growing attention owing to their ability to improve safety and energy density. Among these, halide-based solid electrolytes have emerged as promising candidates because of their high ionic conductivity and stability. We present an MLIP training database for Li 3 MX 6 -type halide electrolytes, where M can be group-3, group-13, and group-15 elements as well as lanthanides, and X represents halogens. The dataset covers a broad chemical space generated by varying crystallographic space groups, introducing isovalent and aliovalent doping, and including anti-site defects. It contains approximately 400k density functional theory (DFT) calculated structures with energies, forces, and stresses, together with hierarchical metadata, providing a resource to accelerate the discovery and optimization of halide electrolytes. We demonstrate the utility of the database by fine-tuning the MACE-MP-0b2-large foundation model: the energy root-mean-square error (RMSE) decreased from 154.8 to 10.0 meV atom -1 and the all-force-component RMSE from 168.3 to 39.0 meV Å-1 on a held-out test set, confirming that domain-specific training data is necessary for structurally disordered halide configurations. ## Early-Enrichment Hit Discovery via Reversible-Work c(t) Estimation in Metadynamics (CTMD) - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-10T00:00:00+00:00 - Categories: Docking & Screening, Targets & Structures, Free Energy & MD - Authors: Venkata Sai Sreyas Adury, Pratyush Tiwary, Xinyu Gu, Mrinal Shekhar - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c01439 - Keywords: Boltz 2, Virtual screening, receptor - Source URL: - Dashboard article: Abstract: Virtual screening for small-molecule binders is often limited by false positives from approximate scoring functions and rigid-receptor assumptions. These can be addressed downstream through accurate but expensive free-energy calculations. At the same time, recent artificial-intelligence-based co-folding methods have been proposed that claim to achieve the accuracy of free-energy methods at much lower cost, but these have not yet delivered consistent improvements in early enrichment and can be confounded by memorization. Here we address this gap by introducing c(t)-based metadynamics (CTMD), a physics-based, high-throughput hit-triaging protocol tailored for early enrichment. CTMD uses the nonequilibrium reversible-work estimator c(t) introduced by Tiwary and Parrinello (Journal of Physical Chemistry B, 2015, 119, 736), computed from a small number of short, independent well-tempered metadynamics trajectories, to rank binding stability without requiring converged binding free energies. We demonstrate that CTMD provides robust early enrichment across diverse targets and chemotypes, while remaining fast and transferable with minimal parameter tuning and resistant to memorization-driven artifacts─underscoring both an immediately deployable physics-based alternative for screening. For these systems, we show how co-folding, particularly Boltz-2, achieves enrichment directly proportional to similarity with the training set and, more worryingly, reproduces this even in the presence of significant modifications to the active site. Given its simplicity of implementation, CTMD should thus be an “embarrassingly″ open-source, early enrichment method available for use by the broad pharmacological and academic community that sits right between approximate but fast docking or AI-based co-folding methods and more expensive but accurate free-energy calculations, and is expected to save significant financial and human capital in drug discovery campaigns. ## Extending AIMNet2 to Macrocyclic Peptides Through Data-Efficient Continual Training - Source: ChemRxiv (preprints) - Date: 2026-09-10T00:00:00Z - Categories: ML Potentials - Authors: Runtian Gao, Roman Zubatyuk, Olexandr Isayev - DOI: 10.26434/chemrxiv.15002334/v2 - External ID: 10.26434/chemrxiv.15002334/v2 - Keywords: MLIP, MACE, DFT - Source URL: - Dashboard article: Abstract: Macrocyclic peptides are among the most promising therapeutic modalities, yet no general-purpose machine learned interatomic potential (MLIP) has been validated for their conformational analysis. We show that AIMNet2 matches the accuracy of MACE-OFF23(S) on the MPCONF196 macrocyclic benchmark, despite macrocycles being entirely absent from its training data, at three-to sixfold lower computational cost. To close the remaining accuracy gap, we introduce a continual training protocol that integrates fewer than 25,000 domain-specific conformers (<0.1% of the training corpus) while preserving accuracy on non-macrocyclic molecules, reducing macrocyclic conformational energy RMSE by 25–27%. Applied to cyclosporine A as an independent test case, the continually trained model reproduces DFT conformational energies with sub-kcal mol −1 accuracy (RMSE 0.78 kcal mol −1 ) and correct rank ordering of all metastable conformers, outperforming all other tested MLIPs and semi-empirical methods. This continual training strategy, together with expanded non-covalent interaction coverage, has been adopted in the AIMNet2 (2025) release, which unifies small-molecule, macrocyclic, and intermolecular interaction treatment in a single production potential. ## Formation of Strong Brønsted Acid Sites on Aluminosilicate Surfaces during Catalytic Cracking - Source: ACS Catalysis (journals) - Date: 2026-09-10T00:00:00+00:00 - Authors: Kaustubh J. Sawant, David Stockwell, Anthony Debellis, Roel Sanchez-Carrera, Lucas Dorazio, Philippe Sautet - Journal: ACS Catalysis - DOI: 10.1021/acscatal.6c03544 - Keywords: equivariant, density functional theory, DFT - Source URL: - Dashboard article: Abstract: Amorphous aluminosilicates are essential components of fluid catalytic cracking (FCC) catalysts, where they provide structural support, hierarchical porosity, and acid functionality within the mesoporous matrix. A molecular-level description of the active acid site ensemble remains challenging because these materials are compositionally heterogeneous and structurally disordered. Most experimental and theoretical studies have focused on Si-doped γ–Al2O3 that serves as a convenient model system. Here, we move beyond this simplified representation toward more realistic and chemically complex aluminosilicate structures. We combine density functional theory (DFT), equivariant machine-learning interatomic potentials (MLIPs), and grand-canonical basin hopping (GCBH) to map structure-acidity relationships for aluminosilicate compositions that arise under FCC-relevant hydrothermal conditions (T = 550 °C and PH2O = 1.2 bar), representative of the reactor inlet. We study a family of mullite surfaces across Al2O3:SiO2 ratios (denoted n:m) and associated oxygen-vacancy contents, and we explicitly model additional SiO2 deposition to emulate silica redistribution during phase evolution from kaolin through spinel intermediates to mullite. Machine-learning potentials trained on structurally related sillimanite surfaces accurately describe Al–Si–O–H chemistry and transfer to vacancy-rich and silica-modified mullite phases. Using NH3 binding as a descriptor of Brønsted acidity, we find that mullite can host acid sites with strengths reaching ΔENH3 ≈ –140 kJ mol−1 in 3:2 mullite, which is comparable to external zeolitic acid sites and substantially stronger than those reported on Si-doped γ–Al2O3. Silica grafting onto otherwise weakly acidic mullite terminations further generates an ensemble of strong bridging and pseudo-bridging Brønsted acid sites. We attribute this enhanced acidity to the local emergence of tetrahedrally coordinated AlIV environments that strongly polarize Al–OH–Si linkages. These results identify defect-rich as well as silica-decorated mullite surfaces as a realistic and potentially dominant source of strong matrix acidity in FCC catalysts under operating conditions. ## Gradient-Based Curvature-Only Time-Derivative Coupling for Surface Hopping and Coherent Switching with Decay of Mixing: Application to SO2 and Molecular Tully Models - Source: Journal of Chemical Theory and Computation (journals) - Date: 2026-09-10T00:00:00+00:00 - Authors: Feven-Alemu Korsaye, Leticia González - Journal: Journal of Chemical Theory and Computation - DOI: 10.1021/acs.jctc.6c01219 - Keywords: TDC - Source URL: - Dashboard article: Abstract: Time-derivative couplings (TDCs) are central ingredients in mixed quantum–classical simulations of nonadiabatic dynamics, yet their evaluation remains a key practical challenge when analytic nonadiabatic coupling vectors or wave function overlaps are unavailable. Curvature-driven approximations provide an appealing alternative by evaluating TDCs from adiabatic potential energies or gradients along nuclear trajectories, motivated by the one-dimensional Baeck–An model. In this work, we examine curvature-based TDC approximations within trajectory surface hopping (TSH) and coherent switching with decay of mixing (CSDM), and introduce a modified gradient-based curvature-only formulation that retains the directional-curvature contribution to the TDC while omitting the explicit acceleration-dependent term at all times in the coupling evaluation. The performance of energy-based, gradient-based, and curvature-only TDCs is evaluated for four benchmark systems of increasing dynamical complexity: ethylene, DMABN, fulvene, and SO2. For ethylene, all curvature-driven schemes closely reproduce reference dynamics obtained with analytic nonadiabatic couplings. For DMABN and fulvene, which involve repeated nonadiabatic passages and back-transfer, the curvature-only approximation yields improved agreement with reference populations by reducing overpopulation effects observed with standard curvature formulations. In the multistate singlet–triplet dynamics of SO2, standard curvature-based TDCs introduce spurious population transfer to the ground state, an artifact that is significantly decreased by the curvature-only scheme in both TSH and CSDM. Overall, the results show that removing the acceleration-dependent contribution leads to a more robust curvature-driven TDC approximation across a wide range of nonadiabatic regimes, while retaining the conceptual simplicity and efficiency of curvature-based approximations. ## Gradient-based Optimization for mRNA Sequence Design - Source: Bioinformatics (journals) - Date: 2026-09-10T00:00:00+00:00 - Authors: Hongmin Li, Goro Terai, Takumi Otagaki, Kiyoshi Asai - Journal: Bioinformatics - DOI: 10.1093/bioinformatics/btag667 - Source URL: - Dashboard article: - Code: Abstract: Motivation Designing mRNA coding sequences that simultaneously optimize RNA accessibility in the translation initiation region and codon adaptation while preserving the encoded protein requires navigating a vast discrete combinatorial space. The inherently discrete nature of codon choices prevents direct application of gradient-based optimization, despite the availability of accurate deep learning predictors such as DeepRaccess for RNA accessibility prediction. Results We present the Input Data Differentiable Designer (ID3), a unified framework for mRNA codon optimization. ID3 treats trained models as fixed differentiable functions and optimizes input data through continuous probability distributions while preserving the encoded amino acid sequence through three constraint mechanisms. The framework shows strong performance in both accessibility optimization and joint accessibility-CAI optimization across diverse protein targets. We also provide convergence analyses from the perspective of trained model input optimization. Availability and implementation Code, datasets, and reproduction scripts are available at https://github.com/Li-Hongmin/ID3.git and archived on Zenodo (DOI: 10.5281/zenodo.18917770). ## Language models as QSPR predictors: Unleashing potential through in-context learning and instruction tuning - Source: ENGINEERING Energy (journals) - Date: 2026-09-10T00:00:00Z - Categories: Property Prediction - Journal: ENGINEERING Energy - DOI: 10.1007/s11708-026-1076-y - External ID: 04982feb8a3109684e4d8ebde77aa0998437175d - Keywords: QSPR - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Leveraging neural network models for drug repurposing: a case study on cardiac hypertrophy - Source: Scientific Reports (journals) - Date: 2026-09-10T00:00:00+00:00 - Authors: Rasmus Magnusson, Markus Johansson, Sepideh Hagvall, Jane Synnergren - Journal: Scientific Reports - DOI: 10.1038/s41598-026-68286-z - Source URL: - Dashboard article: Abstract: Drug repurposing has emerged as an attractive strategy in contemporary pharmaceutical research, presenting an opportunity to expedite drug discovery, minimize developmental costs, and mitigate risks associated with developing new pharmaceuticals. In this study, we investigated a novel approach based on deep learning of human transcriptomic mechanisms for systematic identification of additional therapeutic potential in preexisting drugs. We trained a composite feedforward neural network model using gene expression data sourced from the ARCHS4 compilation of the GEO, encompassing extensive human datasets. Subsequently, disease-associated gene expression data were generated from our stem cell-derived in vitro model of cardiac hypertrophy induced by Endothelin-1 stimulation. These data were employed to identify latent variables associated with genes showing differential expression due to Endothelin-1 stimulation. By examining the differential expression profiles within the model's latent space, we successfully correlated the disease signal with known drug targets found in pharmaceutical compounds cataloged in DrugBank. The model accurately encoded additional disease-related genes beyond the curated gene set, demonstrating its ability to generalize disease associations. Leveraging the model, we identified potential drug candidates, such as lapatinib and amiodarone showing promise in mitigating proBNP concentration associated with cardiac hypertrophy. This study demonstrates the power of deep learning of human transcriptomic mechanisms in swiftly identifying new therapeutic potentials for existing drugs, highlighting the pivotal role of artificial intelligence technologies in accelerating drug development for other complex medical conditions. ## Machine-Learned Potentials for Accelerated CCSD(T)-level Evaluations: Applications to OH(H2O)n - Source: ChemRxiv (preprints) - Date: 2026-09-10T00:00:00Z - Authors: Greta M. Jacobson, Chenrui Shao, Tanya I. Bakalov, Lixue Cheng, Anne B. McCoy - DOI: 10.26434/chemrxiv.15008636/v1 - External ID: 10.26434/chemrxiv.15008636/v1 - Keywords: graph neural network, equivariant - Source URL: - Dashboard article: Abstract: An end-to-end machine learning workflow for the development of potential energy surfaces at the DLPNO-CCSD(T)/aug-cc-pVTZ level is described and applied to hydroxide-water complexes with two to five water molecules. This workflow utilizes a two-part machine learning process, achieving both accuracy and efficiency by combining a molecular orbital-based machine learning (MOB-ML) model with a GPU-accelerated equivariant graph neural network (EGNN) model. For both machine learning models trained in this workflow, the training data is obtained through the combination of small diffusion Monte Carlo (DMC) simulations initiated at each of the optimized structures of interest for a given system size. A farthest point sampling algorithm based on a principal component analysis is employed to generate a diverse set of geometries that spans configuration space, including transition states. By transforming Cartesian coordinates into a graph-based representation, the EGNN model eliminates additional preprocessing of molecular configurations, provides an unbiased treatment of multiple isomers, and allows for isomerization processes among the isomers. An error analysis is performed on the final EGNN models, comparing the potential energies predicted for optimized structures, transition states, and structures that are obtained from snapshots of the ground state wave function sampled by large-scale DMC simulations to those evaluated at the DLPNO-CCSD(T)/aug-ccpVTZ level of theory and basis. Finally, production-run DMC simulations are performed using the validated potentials for both the OH(H 2 O) 2-5 and OD(D 2 O) 2-5 systems, and an analysis of the relative anharmonic zero-point energies is performed. It is found that deuteration does not change the relative energy ordering of the isomers of the OH(H 2 O) 3-5 complexes. Evidence of proton delocalization is identified in the prism isomer of OH(H 2 O) 5 in which a hydrogen atom in one of the solvating water molecules is found to be, on average, equidistant from the oxygen atoms in the hydroxide ion and the hydrogen-bonded water molecule. This delocalization is found to be retained with full deuteration of this complex. ## Mapping high resolution, multidimensional phase diagrams of near-physiological protein condensates - Source: Nature Communications (journals) - Date: 2026-09-10T00:00:00+00:00 - Authors: Tanushree Agarwal, Tomas Sneideris, Fabian Svara, Klavs Jermakovs, Helena Coyle, Seema Qamar, Emanuel Kava, Rob Scrutton, Nicole Pleschka, Priyanka Peres, Gea Cereghetti, Ewa Andrzejewska, Alejandro Diaz-Barreiro, Gaby Palmer, Antonio J. Costa-Filho, Georg Krainer, Tuomas PJ Knowles, Jonathon Nixon-Abell - Journal: Nature Communications - DOI: 10.1038/s41467-026-77696-6 - Source URL: - Dashboard article: Abstract: Biomolecular condensates are membraneless compartments, crucial for organising and regulating diverse cellular processes. Current approaches to study condensate biology either use simplified recombinant protein systems with limited physiological relevance, or complex live-cell models with restricted experimental control and scalability. Here, we present ExVivo PhaseScan, a droplet microfluidics platform that couples mammalian lysate-based reconstitution with scalable analysis to generate high-resolution phase diagrams of compositionally complex protein condensates. We apply this approach to study two multicomponent condensate systems, stress granules and nucleoli, and dissect the physicochemical interactions that influence their stability. We further developed a machine learning pipeline to analyse condensate morphology which we use to reveal how mutations in the amyotrophic lateral sclerosis (ALS)-linked protein Fused in Sarcoma (FUS) remodels condensate properties. We identify liquid-to-solid transitions of mutant FUS within stress granules and nucleoli, and show that these transitions can be reversed by RNA aptamer-based interventions. Together, these findings establish ExVivo PhaseScan as a versatile tool for dissecting the physicochemical and pathological regulation of condensates, with potential to inform therapeutic strategies for diseases driven by aberrant phase transitions. ## ML26: New Ingredients for an Integral-Features Density Functional with Broad Chemical Accuracy - Source: ChemRxiv (preprints) - Date: 2026-09-10T00:00:00Z - Authors: Dayou Zhang, Yinan Shu, Benjamin G. Janesko, Donald G. Truhlar - DOI: 10.26434/chemrxiv.15008672/v1 - External ID: 10.26434/chemrxiv.15008672/v1 - 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 datasets, ML26@MN15 yields the lowest averaged error 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. ## Molecular Origins of Site-Specific CO2 Absorption in Hydrated Deep Eutectic Solvents - Source: ChemRxiv (preprints) - Date: 2026-09-10T00:00:00Z - Categories: Free Energy & MD - Authors: Deepak Kumar Panda, Soumik Das, Salman A. Khan - DOI: 10.26434/chemrxiv.15008638/v1 - External ID: 10.26434/chemrxiv.15008638/v1 - Keywords: free energy perturbation, free energy - Source URL: - Dashboard article: Abstract: Deep eutectic solvents (DESs) have emerged as promising candidates for CO 2 capture because of their highly tunable structures and CO 2 absorption capacities. However, their behavior under practical operating conditions remains poorly understood. In particular, the effect of water on DES structure and their CO 2 absorption capacity remains unclear. In this work, we use free energy perturbation (FEP) and molecular simulations to determine the CO 2 absorption capacities and transport properties of hydrated hydrophilic and hydrophobic DESs. Our results show that increasing water concentration decreases viscosity but also makes CO 2 solvation less favorable in both DESs, revealing a trade-off between transport properties and CO 2 absorption capacity. This trade-off is more favorable for the hydrophobic DES, which shows a substantial reduction in viscosity while maintaining a negative CO 2 solvation free energy. In contrast, the solvation free energy of the hydrophilic DES increases more substantially, becoming positive at 15 wt% H 2 O. Structural analysis reveals that these differences arise from distinct hydration and CO 2 solvation mechanisms in the two DESs. In the hydrophilic DES, CO 2 preferentially localizes around the Br – anion under dry conditions. Upon hydration, water blocks the anionic site, and the preferred CO 2 solvation site shifts to the cation. ## Phosphorylated Analogs of Paracetamol and Their Anti‐Inflammatory Potential - Source: ChemMedChem (journals) - Date: 2026-09-10T00:00:00+00:00 - Categories: Docking & Screening, ADMET & Safety - Authors: Osvaldo León de la Cruz, Gabriel Alfonso Gutiérrez‐Rebolledo, Carlos Zepactonal Gómez Castro, Ángel Daniel Campos Juárez, Porfirio Alonso Ruiz‐Hurtado, Josué Rodríguez‐Lozada, José Luis Castrejón‐Flores, María Guadalupe Ramírez‐Sotelo, Marco Franco‐Pérez, Angel Zamudio‐Medina - Journal: ChemMedChem - DOI: 10.1002/cmdc.70469 - Keywords: Molecular docking, Lipinski, hERG, ADME - Source URL: - Dashboard article: Abstract: Paracetamol is a widely used analgesic and antipyretic agent; however, its use is limited by minimal anti‐inflammatory activity and the risk of hepatotoxicity from prolonged oral use or in acute overdose. To address these limitations, four novel phosphorylated paracetamol analogs (2a–2d) were synthesized via a UV‐radical methodology and evaluated through integrated in silico, in vitro, and in vivo for anti‐inflammatory dermal approaches. Molecular docking suggested plausible binding interactions with COX‐1 and COX‐2 for all analogs. In vitro cytotoxicity assays in THP‐1 cells showed that 2a and 2b did not affect cell viability at 100 μM over 24 h. In a TPA‐induced acute ear edema model in CD1 male mice, 2a produced about 50% inhibition of edema at the lowest tested quantity (0.5 mg/ear), representing a fourfold potency advantage over indomethacin at the same amount, while 2b displayed significant anti‐inflammatory and vasoregulatory activity at higher quantities. Computational ADME profiling indicated that all analogs satisfy Lipinski’s drug‐likeness criteria and exhibit low predicted hERG channel risk. These results support N‐phosphorylation of 4‐aminophenol as a viable strategy to improve the topical anti‐inflammatory efficacy of paracetamol analogs. ## PolyReader: an agentic LLM pipeline for extracting structured data from unstructured polymer scientific literature - Source: ChemRxiv (preprints) - Date: 2026-09-10T00:00:00Z - Categories: LLMs & Agents - Authors: Yuwei Shi, Xinyan Zhong, Yuqi Wei, Heechae Choi, Tianhang Zhou, Zhenghao Wu - DOI: 10.26434/chemrxiv.15008614/v1 - External ID: 10.26434/chemrxiv.15008614/v1 - Keywords: LLM - Source URL: - Dashboard article: Abstract: Machine learning for polymer design requires structured, sample-level records linking polymer identity, synthesis, composition, molecular-weight characteristics, processing conditions, and measured properties. Yet decades of polymer data remain dispersed across the texts, tables, figures, and supporting information of the literature, where individual values are often separated from the samples and experimental context they describe. Here we introduce PolyReader, an LLM agent for provenance-aware extraction of polymer data from scientific papers. Rather than treating a paper as a single block of text, PolyReader represents the manuscript and supporting information as navigable hierarchical section trees. The agent surveys this structure, selectively reads relevant pages, follows cross-references between documents, and integrates dispersed evidence into schema-validated, per-sample records. The resulting extracted fields are thus linked to its source document and page. On a benchmark of 42 polymer papers evaluated by trained annotators, PolyReader achieved F1 scores of 0.92–0.96 across 11 property categories, with comparable performance across three LLM models. Holding the LLM model, parsed document representation, and output schema fixed, PolyReader improved F1 by 13–17 points over single-pass prompting. These results show that reliable polymer-data extraction depends not only on recognizing individual values, but on reconstructing the sample-level relationships distributed throughout a paper. PolyReader produces records that are directly auditable against the primary literature, providing a scalable foundation for constructing traceable polymer datasets for machine learning. The extraction pipeline and benchmark annotations are released for reuse in the community. ## Seamless QM/MM Simulations via a GROMACS-CP2K Interface - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-10T00:00:00+00:00 - Categories: Free Energy & MD - Authors: Dmitry Morozov, Christian Blau, Ole Schütt, Vladimir Mironov, Arno Proeme, Holly Judge, Thomas D. Kühne, Berk Hess, Gerrit Groenhof - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c02073 - Source URL: - Dashboard article: Abstract: Understanding and controlling chemical reactivity in biological systems require atomic-level insight into processes that are often inaccessible to experiments. Hybrid quantum mechanics/molecular mechanics (QM/MM) simulations provide a powerful framework for describing chemical reactions in complex environments, but their practical application remains limited by fragmented software ecosystems, restricted accessibility, and methodological approximations that can compromise accuracy and reproducibility. In particular, many existing QM/MM implementations rely on ad hoc couplings, proprietary software, or truncated treatments of long-range electrostatic interactions. Here, we present a robust and fully periodic QM/MM interface between the open-source molecular dynamics engine GROMACS and the electronic structure theory code CP2K. The implementation enables efficient and reproducible QM/MM molecular dynamics and enhanced sampling simulations with a consistent treatment of long-range electrostatics under periodic boundary conditions. By combining the strengths of two widely used community codes, this interface provides a general and scalable platform for studying chemical reactivity in biological systems and establishes a transparent reference implementation for QM/MM simulations. ## scDEFT: A deep learning framework for drug-effect prediction and counterfactual reasoning - Source: arXiv (preprints) - Date: 2026-09-09T21:02:21Z - Authors: Murthy Devarakonda - External ID: 2609.10831v1 - Source URL: - Dashboard article: - PDF: Abstract: Longitudinal single cell atlases now capture matched pre treatment and post treatment states from responders and non responders, presenting an opportunity to mechanistically explain why two patients on the same drug diverge. We introduce scDEFT (single cell Drug EFfect Transducer), which treats a drug as a conditioning operator on cell representations, enabling prediction and explanation. In scDEFT, feature wise linear modulation produces drug conditioned cell latents, learned under abundant per cell supervision and then frozen. Two independent heads aggregate those latents over shared transcriptional neighborhoods to predict drug induced state change and responder status. A backward stage ranks the latent dimensions by how strongly they separate responders from non responders and maps them to genes under a cell composition control. On a harmonized inflammatory bowel disease atlas of 1.16 million cells, three cohorts and two drug classes, scDEFT predicts state change at 45% of the baseline to reproducibility ceiling headroom and stratifies responders before treatment at AUROC 0.70, where standard predictors remain at chance. These predictions and the drivers behind them support target and co target nomination, patient stratification, and counterfactual prediction of unseen drug cohort effects. ## VMD not for Bio, but for CompChem - Source: Dr. Joaquin Barroso's Blog (feeds) - Date: 2026-09-09T14:59:00+00:00 - Categories: Blog - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Understanding Machine Learning Models Trained on DNA-Encoded Libraries for Virtual Screening - Source: Digital Discovery (journals) - Date: 2026-09-09T14:43:30Z - Categories: Library Design - Authors: Artur Menzeleev, Sathya Chitturi, Geraint Davies, Tony Schroeder, Alpha Lee - Journal: Digital Discovery - DOI: 10.1039/d6dd00162a - Keywords: Virtual Screening, DNA Encoded Libraries, DNA encoded library - Source URL: - Dashboard article: Abstract: DNA-encoded library (DEL) screening enables the rapid generation of billion-scale structure-activity datasets and has become a powerful approach for identifying chemical starting points against challenging biological targets. Recent work has... ## Exploring Transition Metal Complexes with Large Language Models - Source: Chemical Science (journals) - Date: 2026-09-09T14:43:21Z - Categories: Cheminformatics, LLMs & Agents - Authors: Yunsheng Liu, Jacob Toney, Joseph Cavanagh, Kunyang Sun, Andrew Smith Smith, Chung-Yueh Yuan, Roland Gerard St. Michel, Paul A Graggs, F. Dean Toste, Heather Kulik, Teresa Head-Gordon - Journal: Chemical Science - DOI: 10.1039/d6sc00436a - Keywords: SMILES - Source URL: - Dashboard article: Abstract: Here we use a large language model for the exploration of chemical space of transition metal complexes (TMCs), fine-tuning the open-source Llama-3.2-1B model with a variation of the SMILES string... ## Predicting the Toxicity of Chemical Compounds via Hyperdimensional Computing - Source: Molecular Informatics (journals) - Date: 2026-09-09T14:40:36Z - Categories: Cheminformatics - Authors: Fabio Cumbo, Kabir Dhillon, Jayadev Joshi, Bryan Raubenolt, Davide Chicco, Sercan Aygun, Daniel Blankenberg - Journal: Molecular Informatics - DOI: 10.1002/minf.70052 - Keywords: SMILES, Tox21, cheminformatics, property prediction - Source URL: - Dashboard article: Abstract: Accurately and efficiently assessing the potential toxicity of chemical compounds is critical given their wide application across pharmaceutical, industrial, and environmental domains. Traditional toxicological evaluations, which predominantly rely on intensive in vitro and in vivo assays, are frequently slow and expensive. Here, we introduce a novel application of hyperdimensional computing (HDC), a recently developed computational paradigm inspired by the way the human brain works in encoding information, for the efficient classification of chemical compounds as either toxic or nontoxic. Our methodology employs Simplified Molecular Input Line Entry System (SMILES) representations of compounds, drawing data from the comprehensive Tox21 dataset. We delineate a pipeline wherein these chemical structures are encoded into high‐dimensional binary vectors, which subsequently serve as the foundation for training and classification within the HDC framework. This approach leverages HDC's inherent advantages, including its resilience to noise, parallel processing capabilities, and efficacy in identifying intricate patterns. This work demonstrates the viability of HDC as a computationally lightweight first‐pass solution for preliminary toxicity screening. This research significantly contributes to the field of cheminformatics by validating HDC's potential in chemical property prediction, thereby facilitating accelerated identification of hazardous substances and mitigating the reliance on intensive laboratory experimentation. ## Are You Learning Biological Signal or Shortcuts? Auditing and Mitigating Bias in Protein-Protein Interaction Datasets - Source: arXiv (preprints) - Date: 2026-09-09T14:02:18Z - Authors: Judith Bernett, Anton Spannagl, Joel Ås, Markus List, David B. Blumenthal - External ID: 2609.10193v1 - Source URL: - Dashboard article: - PDF: - Code: Abstract: Protein-protein interaction (PPI) databases do not faithfully reflect biological realities. Instead, they are influenced by study and technical biases that distort certain protein and interaction attributes. Machine learning models can exploit these as learning shortcuts if the negative dataset is not constructed with care. So far, the shortcuts introduced during PPI dataset construction have only been examined in isolation. Here, we systematically characterize both reported and, to our knowledge, previously unreported biases in PPI datasets that lead machine learning models to learn shortcuts instead of biological signal. We analyze HIPPIE, IntAct, and STRING, dedicated PPI databases, as well as two datasets derived from 3D-structural information in the Protein Data Bank (PDB). We show that random data splitting introduces strong topological shortcuts. When train-test protein overlap is removed, the resulting datasets still retain usable shortcuts stemming from self-interactions, taxonomic identity, and functional relatedness, whose prevalence interestingly depends on the data source. We further show that sampling negatives from a set of high-confidence non-interactors, an intuitively appealing choice, can amplify the shortcut stemming from functional relatedness. To detect and mitigate these biases, we provide an open Nextflow pipeline that combines similarity-aware, data-loss-minimizing dataset splitting with bias-minimizing negative sampling, both formulated as integer linear programs. Its key concept of quantifying biases to minimize them through optimization-based negative sampling can, in principle, be extended to any machine learning problem where the pool of negative candidates is much larger than the positives and is thus of interest also beyond PPI prediction. ## ADMET-EvO: a self-evolving scientific agent for sustained research across heterogeneous tasks - Source: arXiv (preprints) - Date: 2026-09-09T13:00:00Z - Categories: Cheminformatics, ADMET & Safety - Authors: Yiling Zhou, Yilin Wang, Jianmin Wang, Heqin Zhu, Zirui Wang, Chang-yu Hiesh, Kejun Ying, Jiaqi Wang, Yuzhi Xu, Tingjun Hou, Odin Zhang - External ID: 2609.10121v1 - Keywords: Therapeutics Data Commons, TDC, ADMET - Source URL: - Dashboard article: - PDF: Abstract: Scientific agents can move beyond automated model building by using accumulated evidence to revise both their questions and experimental strategies. The challenge is sustaining this adaptation across heterogeneous tasks without overfitting decisions to internal validation. Absorption, distribution, metabolism, excretion and toxicity (ADMET) prediction provides a demanding setting across diverse assays, datasets and chemical domains. We therefore developed ADMET-EvO, an evidence-gated agent that formalizes endpoints, generates falsifiable hypotheses and tests interventions across data, feature and model axes. It carries supported, rejected and inconclusive outcomes forward to guide each new cycle. Across the 22-task Therapeutics Data Commons (TDC) ADMET benchmark, ADMET-EvO achieved the highest task-normalized score of 96.77. Evidence-guided selection reduced cumulative fitting time by 72.2% within a predefined non-inferiority margin. It also formalized 43 toxicity-related tasks and constructed endpoint-specific predictors. Together, these results show how ADMET-EvO can accumulate evidence, revise its strategy and expand its research scope over time. ## A Systematic Evaluation of Molecule Generation Models for De Novo Drug Design: From Benchmarks to Practical Insights - Source: arXiv (preprints) - Date: 2026-09-09T12:23:27Z - Categories: Cheminformatics, Property Prediction, Docking & Screening, Design de novo - Authors: Xinrui Xu, Xueer Wang, Dan Luo, Sisi Yuan, Xuan Lin - External ID: 2609.10099v1 - Keywords: Molecule Generation, De Novo Drug Design, virtual screening, Transformer, recurrent neural, RNN, diffusion models, molecular representations, receptor - Source URL: - Dashboard article: - PDF: - Code: Abstract: Molecule generation has emerged as a powerful computational tool for de novo drug design, enabling the exploration of 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 reported performance across commonly used benchmarks and evaluation metrics. We also summarize representative experimentally validated case studies. Looking ahead, we discuss future directions in standardized 3D data, interaction-aware generation, receptor flexibility, and multi-objective 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. ## Computational Data as the Fuel for AI in Chemistry - Source: Chemical Science (journals) - Date: 2026-09-09T11:43:15Z - Authors: Ross James Urquhart, Tell Tuttle - Journal: Chemical Science - DOI: 10.1039/d6sc03659g - Source URL: - Dashboard article: Abstract: The rapid growth of machine learning and generative AI in the chemical sciences has placed increasing demands on the data used to train and evaluate these models. Although algorithmic advances... ## An Explainable Machine Learning Framework for Predicting Blood-Brain Barrier Permeability Using Molecular Descriptors - Source: arXiv (preprints) - Date: 2026-09-09T10:42:56Z - Categories: Cheminformatics, Property Prediction, ADMET & Safety - Authors: Fatemeh Mahmoudi - External ID: 2609.10012v1 - Keywords: TPSA, RDKit, MoleculeNet, cheminformatics, Random Forest, Gradient Boosting, XGBoost, LogP, permeability prediction, Molecular Descriptors - Source URL: - Dashboard article: - PDF: Abstract: Blood-brain barrier (BBB) permeability is a critical determinant in the development of central nervous system therapeutics because it directly influences the ability of drug candidates to reach their target sites within the brain. In this study, an explainable machine learning framework was developed to predict BBB permeability using molecular descriptors generated from the MoleculeNet BBBP dataset with the RDKit cheminformatics toolkit. Fifteen physicochemical descriptors extracted from 2,039 compounds were used to train four supervised machine learning algorithms, including Logistic Regression, Support Vector Machine (SVM), Random Forest, and Extreme Gradient Boosting (XGBoost). Hyperparameter optimization was performed using GridSearchCV, while model interpretability was investigated using SHapley Additive exPlanations (SHAP). Among the evaluated models, the optimized XGBoost classifier achieved the best predictive performance, with an accuracy of 88.97%, a precision of 88.92%, a recall of 97.76%, an F1-score of 93.13%, and a ROC-AUC of 0.9282. Stratified five-fold cross-validation further demonstrated the robustness of the proposed model, yielding a mean ROC-AUC of 0.8982 +/- 0.0130. Feature importance and SHAP analyses consistently identified TPSA, HBD, and LogP as the most influential molecular descriptors governing BBB permeability prediction. Overall, the proposed framework provides an accurate, interpretable, and computationally efficient approach for BBB permeability prediction and may serve as a valuable tool for the early-stage screening of CNS drug candidates. ## Enhancing Biocatalytic Retrosynthesis with a Graph-to-Graph Model - Source: Chemical Science (journals) - Date: 2026-09-09T08:03:49Z - Categories: Reaction Informatics - Authors: Binju Wang, Lina Dong, Lin Yao, Yucheng Yang, Yuxiang Gao, Zhihui Jiang - Journal: Chemical Science - DOI: 10.1039/d6sc04187f - Keywords: Retrosynthesis - Source URL: - Dashboard article: Abstract: Biocatalytic synthesis offers a green and sustainable route for chemical production, yet the rational design of biocatalytic routes remains challenging due to the need to jointly consider reaction feasibility and... ## UnitBoost: Managing Compound LLM Systems with a Merge Operator, Not a Model - Source: arXiv (preprints) - Date: 2026-09-09T07:17:23Z - Categories: LLMs & Agents - Authors: Xing Zhang, Guanghui Wang, Yanwei Cui, Mengdie Flora Wang, Peiyang He - External ID: 2609.09815v1 - Keywords: LLM - Source URL: - Dashboard article: - PDF: Abstract: Compound LLM systems often solve a coordination problem by adding a higher-level LLM. The resulting meta-agent reads workers' outputs, writes the final answer, allocates later calls, and decides when to stop. It is expressive, but it also concentrates three control decisions in an opaque, order-sensitive model call. We ask whether the manager needs to be generative at all. UnitBoost replaces that model with a defined meta-level operator: a task-given unit map turns worker outputs into slot-value proposals, a constrained argmax assembles the output, and the slots left unfilled or unsupported become an explicit residual for the next round. The operator is order-free, records unit provenance, and gives a simple guarantee: without coupling constraints, unit-wise maximization under the same admission score dominates selection of any complete candidate. On three held-out benchmarks, it exceeds the best single candidate chosen with gold labels by 0.060-0.195 absolute task-score points and input-matched generative managers by 0.048-0.076. Replacing only the management step improves six compound-system configurations by 0.013-0.182. Residual-directed rounds raise FanOutQA cell F1 from 0.4778 to 0.5524; matched controls show that the true residual outperforms random targets and ordinary rereading, while a label-free supply signal flags exhaustion after one unproductive round. The same analysis measures three conditions in which no such gain is available (one indivisible unit, unavailable unit identity, and an endpoint that charges for every emitted unit) and quantifies cross-unit coupling as a repair cost. The manager gives up semantic freedom and gains order invariance, unit provenance, and testable failure conditions. ## A Hybrid Statistical Deep Learning Framework for Breast Cancer Survival Prediction Using Covariate Transformation and Simulated Gene Expression Data - Source: Journal of Statistical Theory and Practice (journals) - Date: 2026-09-09T00:00:00Z - Journal: Journal of Statistical Theory and Practice - DOI: 10.1007/s42519-026-00641-9 - External ID: c0f61a5288bca0e257ae48ab084a326a312534a2 - Keywords: molecular features - Source URL: - Dashboard article: Abstract: Accurate prediction of breast cancer survival is crucial for precision oncology and individualized treatment planning. Standard survival models, however, typically fail to capture nonlinear clinical relationships, latent biological heterogeneity, and complex interactions among prognostic factors. To address these limitations, this study introduces a Hybrid Machine Learning Framework that combines adaptive covariate transformation, simulated gene expression generation, cross-modal attention fusion, graph-based patient learning, dynamic multi-expert survival modeling, and Bayesian uncertainty estimation. The framework translates clinical variables into clinically meaningful latent representations of prognosis and creates biologically plausible molecular features to boost prognosis prediction. Experimental results show that the proposed model achieves better C-index, AUC, Precision, Recall, and F1-score (0.927, 0.944, 0.931, 0.924, and 0.927, respectively) and a minimum Integrated Brier Score (0.081). Survival stratification analysis clearly differentiates among the low-, intermediate-, and high-risk patient groups, and ablation analysis results support the contribution of each component of the framework. The results suggest that combining transformed clinical data with synthetic genomic data yields a powerful, interpretable, and uncertainty-aware prediction tool for breast cancer survival, with promising potential for precision oncology applications. ## ActiveFusion: Fused Representations Improve Active Learning for Molecular Property Prediction - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-09T00:00:00+00:00 - Categories: Cheminformatics, Property Prediction - Authors: Nelson Evbarunegbe, Shiyun Wa, Luke Taylor, Anna G. Green - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c01708 - Keywords: molecular fingerprints, RDKit, Chemprop, Transformer, Property Prediction, graph neural network, GNN, chemical space exploration, Molecular Property, molecular representation, molecular representations - Source URL: - Dashboard article: Abstract: Active learning provides an efficient strategy for molecular property prediction by iteratively prioritizing compounds for experimental evaluation. However, the effectiveness of active learning pipelines depends strongly on the choice of molecular representation, and systematic understanding of how representation families affect the active learning process in terms of uncertainty and predictive performance remains limited. In this work, we introduce ActiveFusion, a framework for integrating heterogeneous molecular representations within active learning workflows for molecular property prediction. The framework enables systematic evaluation of physicochemical descriptors, molecular fingerprints, learned graph neural network (GNN) representations, and pretrained Transformer-based representations, as well as feature-level fusion strategies that combine complementary chemical information sources. ActiveFusion evaluates models across four molecular property prediction regression tasks. Across datasets, we demonstrate that feature fusion between learned graph representations and physicochemical descriptors consistently improves prediction performance and discovery (average final iteration R2 of 0.71, 0.67, and 0.54 for our overall best representations Chemprop+RDKit, Chemprop, and RDKit, respectively). We show that exploration-driven acquisition strategies enhance scaffold coverage and promote sampling of structurally novel regions of chemical space, and that model-agnostic acquisition of new compounds based on diversity has strong performance. Notably, both pretrained and finetuned Transformer-based embeddings do not consistently outperform physicochemical features or GNN-learned representations in our setting, highlighting the continued relevance of chemically interpretable features and learned features from supervised, task-specific models for active learning applications in molecular property prediction. Overall, ActiveFusion provides a systematic framework for studying representation-acquisition interactions in molecular discovery with representation fusion capabilities. Our study offers practical guidance for designing active learning pipelines that balance prediction accuracy with chemical space exploration. ## Alchemical Free Energy Perturbation Predicts Relative Binding Affinities of Propofol Analogs and Etomidate Stereoisomers at the GABAAR - Source: ACS Omega (journals) - Date: 2026-09-09T00:00:00+00:00 - Categories: Docking & Screening, Free Energy & MD - Authors: Pritam Kumar Panda, Edward J. Bertaccini - Journal: ACS Omega - DOI: 10.1021/acsomega.6c08745 - Keywords: molecular docking, alchemical free energy, FEP, EC50, free energy perturbation, receptor, CHARMm - Source URL: - Dashboard article: Abstract: General anesthetics such as propofol and etomidate exert their clinical effects primarily through positive allosteric modulation of γ-aminobutyric acid type A receptors (GABAAR). Despite decades of research, the quantitative structure–activity relationships governing anesthetic binding remain incompletely characterized at the atomic level. Here, we employ rigorous alchemical free energy perturbation (FEP) calculations using a CHARMm GPU-accelerated protocol to compute relative binding free energies for a series of 13 propofol analogs at the wild-type GABAAR, as well as for the stereoisomeric pair of etomidate (R-etomidate vs S-etomidate). For propofol analogs at the wild-type receptor, computed relative binding free energies (ΔΔG) correlate with experimentally measured GABA EC50 potentiation values, with propofol (2,6-diisopropylphenol) and disec-butylphenol predicted as the most potent analogs, consistent with experimental data (Pearson r = 0.85, 95% CI 0.50–0.96, p = 0.001; mean of three independent runs). For etomidate, FEP calculations correctly predict that R-etomidate binds approximately 1.8 kcal/mol more favorably than S-etomidate, corresponding to a ∼10-fold difference in potency consistent with the known stereoselective pharmacology of this agent. These results demonstrate that alchemical FEP calculations, implemented through a standardized computational protocol, can quantitatively rank-order anesthetic binding affinities and discriminate stereoisomeric preferences at Cys-loop receptor binding sites, providing a framework for rational design of next-generation anesthetic agents that is more quantitatively robust than simple molecular docking methodologies. ## Atorvastatin Shows Limited Disease-Modifying Effects in an A53T α-Synuclein Mouse Model of Parkinson’s Disease - Source: Neurotoxicity Research (journals) - Date: 2026-09-09T00:00:00Z - Categories: Docking & Screening - Journal: Neurotoxicity Research - DOI: 10.1007/s12640-026-00825-y - External ID: 7dc1daaeab24bcd19aeabe7f865715c65ba20019 - Keywords: molecular docking - Source URL: - Dashboard article: Abstract: Neuroprotective effects of statins in Parkinson’s disease (PD) remain uncertain, and their activity in genetic α-synuclein (αSyn) disease models has been insufficiently characterized. We evaluated atorvastatin (ATO) in a mouse model with nigral overexpression of human A53T-mutant αSyn. Mice received ATO by oral gavage at 10 mg/kg/day for 5 weeks and were assessed using behavioral testing, neuropathological assessment, brain transcriptomics, and molecular docking. ATO inhibited cholesterol biosynthesis-related transcriptional programs and broadly remodeled lipid metabolism-associated networks, but did not lead to functional or histopathological benefit. ATO did not ameliorate motor deficits, restore dopaminergic markers, or reduce αSyn protein levels or pSer129-αSyn immunoreactivity. Transcriptomic analysis further showed that ATO failed to reverse the core disease-associated signature induced by A53T αSyn overexpression and instead increased SNCA mRNA. Targeted RNA-seq and western blot analyses showed no parallel increase in Prkn, Gba1, or Lamp2 transcript abundance or in PARKIN, GBA1, or LAMP2A protein expression. Molecular docking, used here as an exploratory structural comparison, suggested a relatively weak predicted interaction between ATO and αSyn when compared with several other statins and provided supportive context for the lack of efficacy. Overall, our findings indicate limited efficacy of ATO in this αSyn-driven setting and support further comparative evaluation of individual statins across complementary PD models. ## BigMixSolDB: Extraction of a solubility database in solvent mixtures with an uncertainty-quantified large language model-based pipeline - Source: ChemRxiv (preprints) - Date: 2026-09-09T00:00:00Z - Categories: LLMs & Agents - Authors: Andrei Voinea, Anna C.M. Thöni, Elija Veenman, Wilhelm T.S. Huck, Tal Kachman, Mathijs F.J. Mabesoone - DOI: 10.26434/chemrxiv.15001616/v4 - External ID: 10.26434/chemrxiv.15001616/v4 - Keywords: LLMs, LLM - Source URL: - Dashboard article: Abstract: The rapidly expanding body of chemical literature contains physical data scattered across unstructured text and complex tables in thousands of publications. Translating this information into machine-readable formats is essential for training data-driven chemical models. The lack of uncertainty quantification in computationally extracted datasets makes manual extraction and curation still necessary. We here address this challenge and present an uncertainty-quantified data extraction pipeline. Our fully automated pipeline utilizes large language models (LLMs) to extract complex tabular and textual chemical data directly from scientific PDFs. We systematically assess several modern LLMs and document-to-text conversion frameworks. We benchmark the uncertainty in extracted data by comparison of our computationally extracted databases with large, existing databases on solubility that are manually curated. Our pipeline achieves between 93 and 79% of the literature contained data, with near-zero median log deviations for structurally matched rows and residual high-magnitude errors concentrated in a small subset of OCR, table-structure, unit-scaling, or reference-dataset discrepancies. We showcase the utility of our pipeline by deploying it to extract data from over 2,000 articles and generate BigMixSolDB, a comprehensive database on solubility in complex mixtures. BigMixSolDB comprises 280,273 solubility entries spanning single, binary, and ternary solvent systems. Our results demonstrate how integrated LLM-based pipelines can be used for literature data extraction that is comparable with manual curation. We envision that these frameworks can be used for large-scale, accurate extraction of literature data into datasets for machine learning applications ## BONAFIDE: a Python framework for the calculation of local features for atoms and bonds in molecules - Source: Journal of Cheminformatics (journals) - Date: 2026-09-09T00:00:00+00:00 - Categories: Cheminformatics - Authors: Lukas M. Sigmund, Michele Assante, Matthew Ball, Mikhail Kabeshov - Journal: Journal of Cheminformatics - DOI: 10.1186/s13321-026-01283-6 - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Chemical Accuracy for Carbapenem Breakdown by Class A beta-Lactamases Using a Transferable Embedded Machine-Learned Potential - Source: ChemRxiv (preprints) - Date: 2026-09-09T00:00:00Z - Authors: Elliot W. Chan, Michael Beer, Kirill Zinovjev, James Spencer, Marc W. van der Kamp, Adrian J. Mulholland - DOI: 10.26434/chemrxiv.15008538/v1 - External ID: 10.26434/chemrxiv.15008538/v1 - Keywords: enzyme, free energy, molecular dynamics - Source URL: - Dashboard article: Abstract: Bacterial resistance to carbapenems (potent β-lactam antibiotics) is a critical global health threat. The primary cause is β-lactamase enzymes that hydrolyze antibiotics. Previously, carbapenems resisted breakdown, making them ‘last resort’ antibiotics, but enzymes with carbapenemase activity have become widespread. Understanding and predicting the activity of β-lactamases will help in developing antibiotics and inhibitors, and combating antibiotic resistance. Effective prediction requires methods that are both accurate and computationally efficient. Multiscale combined quantum mechanics/molecular mechanics (QM/MM) molecular dynamics simulations of reactions can discriminate between class A β-lactamases (the most widely distributed group) with or without carbapenemase activity, and identify determinants of activity. but face an accuracy/efficiency tradeoff: approximate semiempirical QM methods do not predict reaction barriers accurately, but higher level QM calculations require extensive computer time and resources, making them impractical for rapid predictions. Here, we develop and test a transferable ML/MM framework for the deacylation reaction of the carbapenem meropenem, using the electrostatic machine learning embedding (EMLE) scheme. This model gives free energy barriers to within 1 kcal/mol of values derived from experiment, for four different class A β-lactamases. Crucially, the model was trained only on active site structures from QM/MM simulations of a single enzyme. It did not require retraining to generalize across Class A β-lactamases. It is both faster and more accurate than semi-empirical QM/MM simulations. The ML/MM simulations capture the structural and electrostatic determinants of carbapenemase activity, including oxyanion stabilization and active site electric fields, that differ between enzymes. This work establishes transferable EMLE-based ML/MM simulation as a practical route for accurate, efficient modelling of enzyme-catalyzed reactions. ## Conformal Prediction and Multi-Ensemble Gradient Boosting for Subtype-Selective Binding Affinity Estimation Across Human Adenosine Receptors - Source: ChemRxiv (preprints) - Date: 2026-09-09T00:00:00Z - Categories: Property Prediction - Authors: Utkarsh Patel - DOI: 10.26434/chemrxiv.15008546/v1 - External ID: 10.26434/chemrxiv.15008546/v1 - Keywords: QSAR, ChEMBL, Gradient Boosting, XGBoost, Random Forest, graph neural network, Binding Affinity, bioactivity - Source URL: - Dashboard article: Abstract: Subtype selectivity across human adenosine GPCRs remains an intractable medicinal chemistry challenge. The four orthosteric binding pockets share over 70% sequence homology across transmembrane helices III, V, VI, and VII. Consequently, standard QSAR models fail prospectively. They suffer from systemic scaffold leakage, overestimating held-out affinity while producing point estimates that lack calibrated error bars. We built an open-source, leak-free computational platform to solve both failure modes. The architecture combines XGBoost gradient boosting with MAPIE Jackknife+ cross-conformal prediction, Random Forest, and LightGBM, trained on 9,589 curated ChEMBL v34 and GPCRdb bioactivity records. Under strict Bemis-Murcko scaffold partitioning (N\_train = 6,332; N\_test = 1,583), the ensemble achieved an overall R2 of 0.693 and MAE of 0.390 pChEMBL units. On active compounds alone (N\_test = 3,771, structural decoys removed), accuracy reached an overall R2 of 0.865 and MAE of 0.314. Per-subtype R2 values reached 0.753 for A1, 0.884 for A2A, 0.912 for A2B, and 0.886 for A3. Conformal intervals delivered 85.80% empirical coverage at a 90% nominal confidence level. Uncertainty quartiles scaled monotonically with absolute prediction error. A GINE graph neural network trained on identical scaffold splits managed only R2 = 0.248 overall, demonstrating that curated physicochemical descriptors decisively outperform deep graph convolutions in low-to-medium data regimes. Twenty-fold Y-randomization confirmed genuine structure-activity relationships, with all permuted R2 values falling below zero (p < 0.001). External blind validation on 15 novel GPCRdb ligands yielded a 75% selectivity recall accuracy. TreeSHAP features attributions verified that model decisions follow interpretable electrostatic and steric properties. All source code, curated data splits, model weights, and interactive deployment are publicly available. ## Contrastive Reward Based Reinforcement Learning for Goal Directed 3D Molecular Design - Source: ChemRxiv (preprints) - Date: 2026-09-09T00:00:00Z - Categories: Design de novo - Authors: Bohao Li, Yuedong Yang, Mingyuan Xu, Hongming Chen - DOI: 10.26434/chemrxiv.15008550/v1 - External ID: 10.26434/chemrxiv.15008550/v1 - Keywords: molecular generation, Reinforcement Learning, property optimization - Source URL: - Dashboard article: Abstract: Although three-dimensional flow-matching models have become powerful tools for efficient sampling of high-quality 3D molecular structures, reward-based reinforcement learning for flow-matching models is complicated by the absence of a tractable trajectory likelihood for hybrid molecular states comprising categorical features and continuous coordinates. Existing policy-optimization strategies address this difficulty by stochasticizing the generative process or constructing likelihood surrogates, introducing additional approximations. Here, we present ReCoFlow, a novel trajectorylikelihood-free reinforcement learning framework for reward-contrastive post-training of pretrained molecular flow-matching models. Within ReCoFlow, reward-favored molecules provide positive endpoint supervision for the flow-matching model, whereas reward-disfavored molecules are used to construct reference-guided negative targets. Additionally, an anchor loss to an exponential moving average (EMA) reference model constrains the drift of the fine-tuned model. In benchmarking studies including physicochemical property optimization, three-dimensional shape optimization, and pocket-conditioned 3D molecular generation, ReCoFlow consistently improved the performance of flow-matching models over reward-weighted and surrogate-ratio optimization schemes while maintaining high molecular validity. These results suggest that ReCoFlow could be an effective reinforcement-learning strategy for goal-directed molecular design. ## Data-driven strategies for enzyme engineering: From machine learning to protein language models - Source: ChemRxiv (preprints) - Date: 2026-09-09T00:00:00Z - Authors: Shaozhen Ding, Peng Wu, Shubo Li, Yuxuan Yang, Han Zhu, Heng Yu, Yu Tian, Dachuan Zhang, Qian-Nan Hu - DOI: 10.26434/chemrxiv.15008605/v1 - External ID: 10.26434/chemrxiv.15008605/v1 - Keywords: property prediction, enzyme - Source URL: - Dashboard article: Abstract: Enzymes are key biocatalysts that drive diverse biological processes and industrial applications, offering sustainable alternatives to conventional chemical catalysis. However, the enormous enzyme sequence space and the limited functional annotation have created an urgent demand for computational approaches to accelerate enzyme discovery and engineering. Recent advances in computational methods are accelerating enzyme engineering from empirical trial and error toward predictive design. This review summarizes representative computational tools and predictive models in the enzyme engineering workflow, including (1) enzyme data resources, (2) functional enzyme discovery, (3) enzyme property prediction (e.g., kinetics, optimal temperature and PH, solubility, promiscuity, and specificity), (4) de novo enzyme design, and (5) computationally guided enzyme optimization. We further compare current computational strategies, discuss their strengths and limitations in different engineering scenarios, and highlight emerging trends toward foundation models, generative artificial intelligence, multimodal learning, and closed-loop design–build–test–learn platforms. ## From QSAR to deep learning: an interpretable comprehensive pipeline with a read-across approach for mutagenicity prediction via the Enalos Cloud Platform - Source: RSC Advances (journals) - Date: 2026-09-09T00:00:00Z - Categories: Cheminformatics, Property Prediction, ADMET & Safety - Journal: RSC Advances - DOI: 10.1039/d6ra05786a - External ID: 357c2edd82832d67a5a417e6f14bbe506b26b0a3 - Keywords: virtual screening, molecular fingerprints, QSAR, cheminformatics, XGBoost, convolutional neural, molecular descriptors, molecular representations, Ames, chemicals - Source URL: - Dashboard article: Abstract: Assessing the mutagenicity of chemical compounds is essential for ensuring their safe handling and use, thereby minimizing potential health risks. New approach methodologies (NAMs), including computational approaches, provide non-animal alternatives for testing novel materials and chemicals. This study highlights the potential of in silico NAMs, which can contribute to the development of novel Safe and Sustainable by Design (SSbD) chemicals and substances by identifying potentially hazardous ones at an early stage. Emphasis is given to the mutagenicity prediction based on data from the Ames (bacterial gene mutation) test curating them to consider stereo-specific input whenever necessary. A consensus strategy integrating different chemical representations (molecular fingerprints, 2D and 3D descriptors and molecular graphs) and modelling methods, i.e., Quantitative Structure–Activity Relationship (QSAR), read-across and deep learning models, is employed to predict the mutagenic profile of chemical compounds. In this course, a XGBoost model is developed based on molecular descriptors and a graph convolutional neural networks model to classify compounds as mutagens and non-mutagens based on the Ames test data. The devised read-across methodology is based on a guided-k-Nearest Neighbours scheme (guided-kNN) where two different molecular representations (molecular fingerprints and descriptors) are considered for neighbour selection and predictions generation. The mutagenicity predictions from the three models are integrated in a majority voting scheme to enhance the overall predictive accuracy (83% in external validation) and reduce individual model biases. Interpretation of the descriptors involved in prediction is performed through explainable AI (XAI) methods to provide insight to the mutagenicity mechanism. To enhance the interpretability of the XAI-derived insights and reinforce user confidence in the models' predictions, the involved descriptors are mapped to key events leading to mutations within the Adverse Outcome Pathway (AOP) networks. Apart from the development of reliable and interpretable mutagenicity models, emphasis is given on delivering a pipeline for the generation of 3D descriptors that can be used as the basis for future cheminformatics models. To support transparency and reproducibility of the results of our work, the curated mutagenicity dataset used for modelling is disseminated through the ChemPharos database (https://db.chempharos.eu/datasets/Datasets.zul?datasetID=ds18), the modelling steps are documented following the standardized Modelling Data (MODA) guidelines and the consensus model is freely available via the Enalos Cloud platform (https://www.enaloscloud.novamechanics.com/insight/polis/), to facilitate virtual screening of novel compounds. ## Gut Microbiota and Aldosterone Regulate Natriuretic Peptide B Expression to Drive Mitophagy and Metabolic Reprogramming in Sepsis-Like Model of Myocardial Injury. - Source: Journal of the American Heart Association (journals) - Date: 2026-09-09T00:00:00Z - Categories: Docking & Screening - Journal: Journal of the American Heart Association - DOI: 10.1161/JAHA.125.046120 - External ID: f8fab9aef3ec050a79c5ef5780da067602f83149 - Keywords: Molecular docking - Source URL: - Dashboard article: Abstract: BACKGROUND Myocardial injury is a major contributor to mortality in sepsis, yet the mechanisms underlying gut-heart communication in sepsis-induced myocardial injury remain insufficiently defined. Natriuretic peptide B (NPPB) is a cardiac stress-responsive gene, but its involvement in mitochondrial homeostasis and metabolic regulation is unclear. This study investigated how gut microbiota and aldosterone influence myocardial mitophagy and metabolic reprogramming through NPPB in sepsis-induced myocardial injury. METHODS A sepsis-like myocardial injury model was induced in mice by intraperitoneal lipopolysaccharide (LPS). Fecal microbiota transplantation from septic mice into pseudo-germ-free recipients assessed microbial contributions. Metagenomic, metabolomic, and transcriptomic analyses identified disrupted metabolites and cardiac gene signatures. Heart-specific NPPB-knockout mice were used to determine its in vivo role. Mitochondrial function and metabolic alterations were evaluated by energy metabolism assays. In vitro, aldosterone-treated AC16 cardiomyocytes were used to examine NPPB-mediated mitophagy and metabolic changes. Molecular docking, dynamics simulation, and machine-learning screening identified Lestaurtinib, whose therapeutic effects were validated pharmacologically. RESULTS Sepsis caused pronounced microbial dysbiosis and elevated aldosterone levels. Multi-omics analysis identified NPPB as a central regulator of mitophagy and metabolic remodeling. NPPB deficiency mitigated mitochondrial impairment and metabolic disturbances in vivo. Aldosterone upregulated NPPB in cardiomyocytes, promoting mitophagy and metabolic reprogramming. Lestaurtinib, identified as a candidate targeting the aldosterone-NPPB axis, improved cardiac structure and function while partially restoring microbial and metabolic homeostasis. CONCLUSIONS This study uncovers a novel gut microbiota-aldosterone-NPPB axis driving LPS-induced myocardial injury through dysregulated mitophagy and metabolism and highlights Lestaurtinib as a potential therapeutic strategy for sepsis-induced myocardial injury. ## Harnessing coumarin privilege to achieve potency and selectivity: hypoxia-targeted alkyl coumarin-tethered phenyl quinazolinones as dual hCA IX/HSP90α inhibitors with chemosensitizing activity - Source: RSC Advances (journals) - Date: 2026-09-09T00:00:00Z - Categories: Docking & Screening, ADMET & Safety - Journal: RSC Advances - DOI: 10.1039/d6ra06511b - External ID: 5e16c48144eff6c3c560b6bee11e59866554b273 - Keywords: Molecular docking, KI value, ADME, IC50, molecular dynamics - Source URL: - Dashboard article: Abstract: Hypoxic tumor regions remain challenging to treat in breast cancer therapy due to extracellular acidification, chemoresistance, and other adaptive mechanisms promoted by the coordinated actions of tumor-associated carbonic anhydrase IX (hCA IX), heat shock protein 90 alpha (HSP90α), and other proteins. Inspired by the privileged role of coumarins as selective hCA IX inhibitors and as HSP90α modulators, together with the ATP-mimetic and anticancer properties of phenyl quinazolinones, a series of alkyl coumarin-tethered phenyl quinazolinones (4a–l) were rationally designed and synthesized as hypoxia-targeted dual hCA IX/HSP90α inhibitors. 1H NMR, 13C NMR-DEPTQ, HRMS, elemental analysis, and HPLC unequivocally established the structures of the synthesized compounds. The synthesized hybrids exhibited potent, highly selective inhibition of hCA IX (in the nanomolar range), while displaying negligible activity against the off-target isoforms hCA I and II. Compound 4b emerged as the lead derivative, exhibiting a KI value of 51.7 nM against hCA IX together with exceptional selectivity (>1934-fold over hCA I and hCA II). Remarkably, 4b also exhibited potent HSP90α inhibitory activity (IC50 = 13.21 nM), comparable to that of ganetespib. Furthermore, 4b preferentially suppressed the proliferation of hypoxic MCF-7 and MDA-MB-231 breast cancer cells, sensitized them to doxorubicin, and induced p53-mediated mitochondrial apoptosis. Molecular docking and molecular dynamics simulations against hCA IX and HSP90α, along with in silico ADME and toxicity studies, further supported the favorable binding behavior, stability, and druglike characteristics of the lead compound. ## Local Variational Graph Coarsening for Machine Learning Coarse-Grained Molecular Dynamics Simulations - Source: Journal of Chemical Theory and Computation (journals) - Date: 2026-09-09T00:00:00+00:00 - Authors: Soumya Mondal, Subhanu Halder, Debarchan Basu, Franz Görlich, Sandeep Kumar, Tarak Karmakar - Journal: Journal of Chemical Theory and Computation - DOI: 10.1021/acs.jctc.6c00825 - Keywords: MACE, Message Passing, Molecular Dynamics - Source URL: - Dashboard article: Abstract: Coarse-grained (CG) molecular dynamics (MD) simulations can simulate large molecular complexes over extended time scales by reducing degrees of freedom. A central challenge in CG modeling is the selection of the CG mapping algorithm, which directly influences both accuracy and the interpretability of the model. Despite significant progress, effective strategies for optimal CG mappings remain a challenging task, highlighting the necessity for a comprehensive computational framework. In this work, we introduce a multilevel graph-coarsening framework for coarse graining, inspired by recent advances in graph reduction with spectral and cut guarantees. CG sites are generated through edge contractions based on a local variational cost metric, while preserving essential spectral properties of the original graph. This construction ensures structural fidelity across scales. To learn the corresponding CG energy functions, we utilize the Message Passing Atomic Cluster Expansion (MACE), yielding efficient yet accurate CG potentials. We demonstrate the generality of the framework across isolated molecules, bulk molecular liquid, and crystal. Our approach provides a data-driven bottom-up computational framework for the development of systematically improvable CG potentials with controlled accuracy and physical interpretability. ## Mechanism-Guided Antimicrobial Peptide Design through Membrane-Surface Fingerprinting and Graph Diffusion - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-09T00:00:00+00:00 - Categories: Property Prediction - Authors: Zhenyu Ma, Sixin Tian, Mengmeng Sheng, Hong Cheng, Jingyi Zhu, Chunyi Yang, Limei Xu, Min Xiao, Jian Li, Yanyan Li, Xukai Jiang - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c02664 - Source URL: - Dashboard article: Abstract: Antimicrobial peptides (AMPs) are promising antibiotic alternatives, but current AI-based discovery often relies on sequence labels and lacks explicit modeling of peptide–membrane interactions. We developed Membrane-MaSIF, an MD-derived membrane surface matching model designed to capture peptide–membrane compatibility beyond sequence-level AMP labels. Using LL-37–perturbed Acinetobacter baumannii outer membranes, Membrane-MaSIF represents binding, insertion, and pore-like perturbation states as computable surface fingerprints encoding local geometry and physicochemical features. We further integrated Membrane-MaSIF with PepGraph-Diffusion for de novo AMP candidate generation and applied it to an external SPLUNC1 α4-derived A4 analogue library for known-scaffold prioritization. Experimental validation of selected candidates, including peptide 71 and high-scoring A4 analogues, supported the utility of membrane surface matching for enriching membrane-active antibacterial peptides. This framework provides a mechanism-aware strategy for Gram-negative AMP discovery and optimization. ## Michael addition-based synthesis of lawsone-derived benzopyrans: antibacterial, CREST and ADMET studies - Source: RSC Advances (journals) - Date: 2026-09-09T00:00:00Z - Categories: Docking & Screening, ADMET & Safety - Journal: RSC Advances - DOI: 10.1039/d6ra07021c - External ID: 38b8072b60492e9ae152d2814976252ad8c71ca6 - Keywords: molecular docking, ADMET, pharmacokinetic - Source URL: - Dashboard article: Abstract: Using the Michael addition reaction, we have developed a simple and facile method for the synthesis of novel lawsone-based highly substituted benzopyrans in good to excellent yields (62–94%) at room temperature. All the synthesized compounds were characterized by 1H NMR, 13C NMR, FTIR and HRMS. The relative configuration was assigned by ROESY spectrum, showing 1,3-syn interaction between H1 and H3 protons and further supported by CREST analysis. The Structure-E having SSR configuration possesses the lowest relative electronic energy among the eight possible structures. Following this, the in vitro antibacterial activity was evaluated for the synthesized compounds against two bacterial strains, E. coli (PDB ID: 1KZN) and S. aureus (PDB ID: 3G75), using DNA gyrase as the target. Among the tested compounds, 3p with 6,8-dichloro and –OMe appended phenyl group exhibited the highest bacterial inhibition potency with an MIC value of 6.25 µg mL−1 against E. coli and 12.5 µg mL−1 against S. aureus respectively. Additionally, in silico molecular docking results also revealed that 3p has an excellent binding free-energy estimate of −9.3 kcal mol−1 and −8.2 kcal mol−1 against E. coli and S. aureus. As well, the synthesized compounds with promising pharmacokinetic and drug-like properties were established by ADMET investigations. Consequently, the results of in vitro and in silico assays discovered the capabilities of these compounds as active antibacterial drugs. ## Molecular modeling design of antitrypanosomal pyrazolone derivatives targeting Chagas disease: QSAR, docking, molecular dynamics and free energy calculations - Source: Journal of Molecular Modeling (journals) - Date: 2026-09-09T00:00:00+00:00 - Categories: Property Prediction, Docking & Screening, ADMET & Safety, Free Energy & MD - Authors: Karen Karla F. Sousa, Ana Beatriz S. Barbosa, Fábio José B. Cardoso, José Rogério A. Silva, Fábio A. Molfetta - Journal: Journal of Molecular Modeling - DOI: 10.1007/s00894-026-06923-0 - Keywords: Molecular docking, AutoDock Vina, QSAR, ADMET, free energy calculations, free energy, GAFF, force fields, molecular dynamics, MD simulations - Source URL: - Dashboard article: Abstract: Context Chagas disease, caused by T. cruzi , remains a neglected tropical disease with limited therapeutic options, underscoring the urgent need for novel antitrypanosomal agents. A computer-aided drug design (CADD) protocol was applied to investigate pyrazolone derivatives with reported antitrypanosomal activity and their putative interaction with cruzain (Cz), a cysteine protease essential for parasite survival. A 2D-QSAR model built from a curated set of 33 phenyl-dihydropyrazolone derivatives guided the design of four new candidates (LMM1–LMM4), which showed intermediate predicted antitrypanosomal activity within the chemical space of the original pyrazolone series. Molecular docking identified favorable binding within the Cz active site, with contacts at residues Ser61, Gly66, and Leu67. MD simulations indicated overall structural stability of the protein–ligand complexes and revealed ligand-dependent conformational behavior within the Cz binding site. MM/GBSA calculations suggested that van der Waals interactions are major contributors to the computed binding energies; however, the MM/GBSA ranking did not fully reproduce the phenotypic antitrypanosomal activity trend. Per-residue decomposition and free energy landscape analyses were therefore interpreted qualitatively, suggesting that contacts with S2/S3 subsite residues and conformational adaptability may contribute to Cz recognition but do not, alone, explain whole-cell potency. These findings support pyrazolones as antitrypanosomal scaffolds and provide structural hypotheses for future experimental validation against Cz. Methods Molecular geometries were optimized at the PM6 level (MOPAC2016); descriptors were calculated with ChemDes and UseGalaxy, selected via OPS and a genetic algorithm (QSARINS), and the QSAR model was built using PLS regression (QSAR modeling software). Molecular docking was performed with GOLD, FITTED, and AutoDock Vina 1.2.3 against Cz (PDB: 3KKU). ADMET properties were predicted using OSIRIS Property Explorer, SwissADME, and ADMET-AI. Partial atomic charges were derived at the HF/6-31G\* level with RESP fitting (Gaussian 09); MD simulations (3 × 250 ns per system) were run with AMBER20 using GAFF/ff14SB force fields and TIP3P solvent. Binding free energies were calculated by MM/GBSA (MMPBSA.py). ## Multifaceted In Silico and DFT-Based Analysis of 2-Aminosuccinic Acid on Ferroptosis-Related Targets - Source: ACS Omega (journals) - Date: 2026-09-09T00:00:00+00:00 - Authors: Mehmet Hanifi Kebiroğlu, Serap Yalcin Azarkan, Nevin Çankaya - Journal: ACS Omega - DOI: 10.1021/acsomega.6c02875 - Keywords: molecular docking, DFT, density functional theory - Source URL: - Dashboard article: Abstract: Metabolic reprogramming and redox imbalances in cancer cells play a significant role in tumor growth and development of treatment resistance. In recent years, ferroptosis, a type of programmed cell death characterized by iron-dependent lipid peroxidation, has emerged as a promising target for cancer therapy. Amino acid transport, glutathione (GSH) homeostasis, and antioxidant defense mechanisms are critically important in the regulation of ferroptosis. In this study, the interactions of 2-aminosuccinic acid, involved in amino acid metabolism, with key molecular targets associated with ferroptosis were investigated using an in silico molecular docking approach. Docking analyses revealed that the compound showed significant binding affinities with SLC1A1 (EAAT3), SLC7A11 (xCT), and the GPX4/GSH system. The observed binding energy of −5.0 kcal/mol with SLC7A11 is noteworthy in terms of modulating intracellular cysteine/glutathione balance and regulating ferroptosis sensitivity. These findings suggest that 2-aminosuccinic acid may be a potential molecule capable of influencing ferroptosis in cancer cells via amino acid transport and redox regulation. In addition, the structural, electronic, spectroscopic, and thermodynamic properties of 2-aminosuccinic acid were extensively investigated using density functional theory (DFT) with the 6-311G basis set. The molecule’s geometry was optimized to a stable conformation, and the calculated bond lengths followed values reported in the literature. The HOMO–LUMO energy gap (ΔE = 3.308 eV) indicated that the molecule has high kinetic stability and low chemical reactivity. Theoretical FT-IR, 1H NMR, and 13C NMR spectra confirmed the functional groups, while UV–vis analyses revealed strong absorptions in the 150–200 nm range due to π→π\* and n→π\* transitions. Molecular electrostatic potential (MEP) maps have shown that oxygen atoms form reactive centers as electron-rich regions. Non-covalent interaction (NCI) and Hirshfeld surface analyses revealed that crystal packing is largely stabilized by O···H and H···O hydrogen bonds. Thermo-chemistry map (TCM) analyses detailed the temperature-dependent behavior of thermodynamic parameters such as heat capacity and entropy in the 300–900 K temperature range. This study presents a holistic approach to both the interactions of 2-aminosuccinic acid with ferroptosis-related molecular targets and its fundamental theoretical properties; the findings suggest that this compound could be a potential candidate for future applications in cancer treatment, ferroptosis-targeted drug design, and medicinal chemistry. ## Multiscale Investigation of Triboelectric Nanogenerator: An Expanded DFT Calculation Based on Machine Learning - Source: ACS Omega (journals) - Date: 2026-09-09T00:00:00+00:00 - Authors: Yongsheng Huang, Mitsuhiro Matsumoto - Journal: ACS Omega - DOI: 10.1021/acsomega.6c06130 - Keywords: DFT, density functional theory, molecular dynamics, MD simulations - Source URL: - Dashboard article: Abstract: Various types of triboelectric nanogenerators (TENGs) have been proposed, which rely on the details of electron transfer, ion migration, and mechanochemistry, but their largely different scales complicate the quantitative analysis of their contributions. In this study, we first investigate the electron transfer behavior at the nanoscale using density functional theory (DFT) calculations for the contact of two typical insulating materials, poly(vinyl chloride) (PVC) and poly(vinyl alcohol) (PVA), with an adsorbed water layer in between. The DFT results are then used to construct machine learning (ML) potentials, which enable us to execute larger-scale classical molecular dynamics (MD) simulations. The results demonstrate that electron transfer contributes to the generation of an electrostatic potential difference among materials, while ion migration compensates for part of the potential, leading to a reduced power output. This approach with ML establishes a multiscale simulation framework that links quantum-level mechanisms based on DFT to the investigation of mesoscale behaviors using classical MD simulations, providing insights into the interplay of charge transfer processes and offering guidance for the design of efficient TENG materials. ## Navigating off-target effects in CRISPR-based genome editing for safer gene therapies - Source: Discover Genetics and Evolution (journals) - Date: 2026-09-09T00:00:00Z - Journal: Discover Genetics and Evolution - DOI: 10.1007/s00294-026-01339-y - External ID: bada72a1bd22a78a673afef7e286217c76397637 - Source URL: - Dashboard article: Abstract: Off-target activity remains a significant challenge in the clinical development of CRISPR-based therapies. As programmable nucleases advance from experimental tools toward approved medicines, the ability to predict, detect, and mitigate unintended genomic editing events has acquired direct translational importance. This review examines the molecular basis of off-target cleavage across three nuclease platforms—zinc finger nucleases (ZFNs), transcription activator-like effector nucleases (TALENs), and CRISPR-Cas systems—with emphasis on mechanistic and structural factors that govern mismatch tolerance, including PAM sampling, seed region thermodynamics, and chromatin accessibility. In silico prediction methods are surveyed from early alignment-based approaches through feature-engineered machine learning models to more recent deep learning architectures, with critical attention to training data limitations, cross-platform generalisability, and the distinction between exhaustive genomic search tools and predictive scoring models. The experimental detection landscape is reviewed with expanded coverage of established and newer unbiased genome-wide assays, including their biological principles, sensitivity characteristics, and suitability for structural variant detection. Particular attention is given to next-generation editing modalities—base editors, prime editors, and Cas12/Cas13 systems—and to epigenome editing approaches using non-cleaving CRISPR platforms, which avoid double-strand breaks but retain specificity concerns. The review also addresses delivery-related off-target risks, including tissue-level biodistribution and germline exposure, alongside mitigation strategies. It closes by outlining directions for improving standardisation, expanding structural variation surveillance, and extending safety characterisation to the full diversity of editing platforms now approaching clinical use. ## Network pharmacology, molecular docking, and experimental validation of the mechanisms of Withaferin a in treating periodontitis - Source: Clinical Oral Investigations (journals) - Date: 2026-09-09T00:00:00Z - Categories: Docking & Screening - Journal: Clinical Oral Investigations - DOI: 10.1007/s00784-026-07120-2 - External ID: 1ec31f13f7f5f334624eb9ef0785b9803ffbd920 - Keywords: molecular docking - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Non-bridging oxygen as the master variable linking structure to the elastic, electronic, and photonic properties of heavy, polarizable cesium silicate glasses - Source: ChemRxiv (preprints) - Date: 2026-09-09T00:00:00Z - Authors: Hicham Jabraoui, Yann Vaills, Michael Nolan - DOI: 10.26434/chemrxiv.15008554/v1 - External ID: 10.26434/chemrxiv.15008554/v1 - Keywords: machine learned force field, force field, ab initio - Source URL: - Dashboard article: Abstract: Cesium silicate glasses, (Cs2O)x(SiO2)1−x, are candidate hosts for optical, dielectric and nuclear-waste-immobilization applications, yet the large, polarizable Cs+ cation is poorly described by conventional potentials. We combine a firstprinciples reference with a single, composition-transferable machine-learned force field for the full range x = 0–0.30, built by pooling the ab initio data of all compositions with no empirical parametrization. First-principles calculations show wide-gap insulators whose gap narrows from 5.9 to 3.3 eV, while the static dielectric constant nearly doubles and the refractive index and optical basicity rise, in agreement with experiment. The pooled potential then scales these cells to nanometre-sized glasses, reproducing the density (within ∼2%, and 5% at x = 0.20), the intact SiO4 network, and the bridging-to-non-bridging-oxygen conversion and Qn depolymerization. Its elastic constants agree with Brillouinscattering experiments, and its pair-correlation functions and structure factors match the ab initio reference. A single structural variable, the non-bridging oxygen, governs these trends quantitatively: the moduli decay exponentially to a stiffness floor with a common non-bridging-oxygen scale and the gap narrows linearly with it, making it a quantitative descriptor linking structure to the elastic, electronic and photonic properties and a transferable basis for designing heavy-alkali silicate glasses. ## Noncarbohydrate Inhibitors of Sialic Acid-Binding Immunomodulatory-Type Lectin-7 (Siglec-7) Discovered from Genetically Encoded Bicyclic Peptide Libraries - Source: Journal of the American Chemical Society (journals) - Date: 2026-09-09T00:00:00+00:00 - Authors: Danial Yazdan, A. Michael Downey, Ana Gimeno, Caishun Li, Edward N. Schmidt, Jeffrey Y. K. Wong, Caleb Loo, Jaesoo Jung, Ryan Qiu, Ewa Lis, Lily Lindmeier, June Ereño-Orbea, Jesús Jiménez-Barbero, Matthew S. Macauley, Ratmir Derda - Journal: Journal of the American Chemical Society - DOI: 10.1021/jacs.6c14068 - Keywords: IC50, enzyme, receptor - Source URL: - Dashboard article: Abstract: Glycan-binding proteins (GBPs) are among the most difficult drug targets, limiting clinical progress against therapeutically important GBPs. We employed bicyclic genetically encoded libraries (BiGELs), produced by the chemical modification of phage-displayed peptide libraries with twofold symmetric linchpins, to discover inhibitors of therapeutically relevant Siglec-7:GD3 interactions. Next-generation sequencing (NGS) analysis of BiGEL panning against Siglec-7 yielded 815 candidates, of which 23 hits yielded KD = 1–100 μM, as determined by surface plasmon resonance (SPR). Competitive enzyme-linked immunosorbent assays (ELISA) identified leads that disrupted the Siglec-7:GD3 interaction with IC50 = 3-300 μM. Machine learning models trained on NGS datasets identified additional inhibitors with equivalent potencies. Alanine scans of 8c (SWCRPATVNC, IC50 = 3.8 μM) and 12c (SFCHYPTHVC, IC50 = 11 μM) identified residues crucial for activity. Ring reshaping studies of 8c highlighted the critical role of bicyclic topology, yielding analog 46e (SAAAAAWCRPATVNC, IC50 = 9.5 μM), which was further evolved into 67e (STVTQHWCRPATVNC). The multivalent display of lead bicycles alongside ∼100 glycans in a Liquid Glycan Array (LiGA) enabled the comparison of binding to Siglec-7-expressing cells. LiGA assays confirmed the binding of the bicycles to Siglec-7 but revealed considerable nonspecific interactions with receptor-negative cells. Saturation transfer difference nuclear magnetic resonance (STD-NMR) spectroscopy revealed that 46e binds to Siglec-7 at a site distinct from the V-Ig domain, suggesting inhibition through an allosteric site. Together, these results demonstrate that BiGEL enables the discovery of bicyclic peptides for undruggable Siglec targets, but highlights future challenges in molecular discoveries that aim to identify small, noncarbohydrate inhibitors of GBPs. ## Ochratoxin A perturbs reproductive cell homeostasis involving PI3K-Akt-related stress signaling: Network toxicology, molecular simulation, single-cell and in vitro evidence. - Source: Environmental toxicology and pharmacology (journals) - Date: 2026-09-09T00:00:00Z - Categories: Docking & Screening - Journal: Environmental toxicology and pharmacology - DOI: 10.1016/j.etap.2026.105168 - External ID: 78665e88962f7602e647dfd91fcc9314f773c4af - Keywords: molecular docking, IC50, molecular dynamics - Source URL: - Dashboard article: Abstract: Ochratoxin A (OTA) is a persistent foodborne environmental contaminant with recognized systemic toxicity, but the molecular basis of its reproductive effects remains incompletely defined. We integrated network toxicology, molecular docking, molecular dynamics simulation and public single-cell RNA sequencing to generate candidate mechanisms, and then evaluated selected predictions using cell-viability and transcript-level assays. Intersections between OTA-related and reproductive-injury targets yielded 318 testicular and 426 ovarian candidate targets. AKT1, MAPK1, PIK3CA, SRC and TP53 were shared network hubs, while enrichment analysis prioritized PI3K-Akt signaling, apoptosis, p53, FoxO and MAPK pathways. Docking predicted favorable OTA poses for the five hub proteins, and 100-ns simulations of the selected AKT1-OTA and PIK3CA-OTA complexes showed persistence of the predicted poses within the modeled systems; these computational findings do not establish biological binding or direct target engagement. Public single-cell data localized the candidate genes to germ, somatic and steroidogenic cell populations. In vitro, OTA reduced 24-h viability with IC50 values of 9.25 micromolar in GC-1 cells and 27.94 micromolar in KGN cells and altered hub-gene transcripts. Taken together, the results suggest that OTA-associated reproductive cell stress may involve PI3K-Akt-related survival signaling and apoptosis-linked responses, but protein-level and functional validation is required. ## Performance of universal machine learning potentials in global optimization of inorganic crystal structures - Source: Machine Learning: Science and Technology (journals) - Date: 2026-09-09T00:00:00+00:00 - Authors: Edan T Marcial, Laxman Chaudhary, Olesya Gorbunova, Aleksey N Kolmogorov - Journal: Machine Learning: Science and Technology - DOI: 10.1088/2632-2153/ae94e1 - Keywords: machine learning potentials, MACE, ab initio - Source URL: - Dashboard article: Abstract: Rapid development of universal machine learning potentials (uMLPs) and expansion of training data sets are reshaping the state of the art in atomistic simulation, highlighting the need for concurrent systematic benchmarking of their capabilities. Global optimization is among the most demanding uMLP applications because unconstrained exploration includes probing motifs not present in reference sets. We examined twelve pretrained uMLPs in unconstrained evolutionary searches to assess whether these models can consistently predict complex nonmagnetic crystal structure ground states under ambient pressure across diverse inorganic systems. Our findings demonstrate that the considered M3GNet, MACE-MATPES, MACE-mh-1, SevenNet, EquiformerV2, EquiformerV3, MatterSim, GRACE, eSEN, Orb-v3, PET-MAD, and PET-OAM models span a wide performance range, from near ab initio to essentially non-predictive, in their ability to resolve competing phases within low-energy basins. Additional tests on hcp-Zn, MB \_4 (M = Cr, Mn, and Fe), and LiB \_\{y\} ( y ≈ 0.9 ) ground states reveal that several uMLPs capture fine energy differences arising from subtle electronic structure features. ## Pharmacophore model guided 3D molecular generation through diffusion model - Source: Journal of Cheminformatics (journals) - Date: 2026-09-09T00:00:00+00:00 - Categories: Docking & Screening, Design de novo - Authors: Bohao Li, Xinyu Wu, Ting Ran, Jingpeng Zhong, Pan He, Yongzhi Lu, Miru Tang, Yuedong Yang, Mingyuan Xu, Jinsai Shang, Hongming Chen - Journal: Journal of Cheminformatics - DOI: 10.1186/s13321-026-01298-z - Keywords: molecular generation, Pharmacophore model, diffusion model - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Piano‐Stool Ru (II) Complexes as Modulators of Human Prion Protein PrP 106–126 Aggregation - Source: ChemMedChem (journals) - Date: 2026-09-09T00:00:00+00:00 - Categories: Docking & Screening - Authors: Rahul Chauhan, Himanshi Kumawat, Sakshi Saini, Abhishek Panwar, Prashant Kukreti, Nandita Medda, Partha Roy, Kaushik Ghosh - Journal: ChemMedChem - DOI: 10.1002/cmdc.70480 - Keywords: molecular docking, binding affinity - Source URL: - Dashboard article: Abstract: Prion diseases are neurodegenerative disorders caused by the accumulation of misfolded prion proteins, leading to neurotoxicity and neuronal death. Creutzfeldt‐Jakob disease (CJD) in humans and Bovine spongiform encephalopathy (BSE) in animals are among the most prominent prion disorders. Therefore, the search for molecules that can serve as drugs for prion diseases has gained significant attention. In our study, we synthesised and characterised two piano‐stool ruthenium complexes, RuBT and RuBI, based on benzazole‐quinoline scaffolds. Molecular structures of RuBT and RuBI were determined by X‐ray crystallography. The anti‐aggregation effects of the complexes on PrP 106–126 aggregation were studied through ThT assay, CD, TEM, and AFM. Seeded aggregation assays demonstrated their ability to modulate seed‐induced fibril formation. The interaction between the complexes and PrP 106–126 was analysed using molecular docking and UV–visible spectroscopy. Molecular docking analysis showed good binding affinity with the PrP 106–126 . To demonstrate bioavailability and transport, the DNA and HSA binding affinities of complexes were computed. The complexes exhibited neuroprotective effects in HT‐22 cells by reducing damage induced by PrP 106–126 . They also showed anticancer activity on human neuroblastoma SH‐SY5Y cells. These findings suggest that these complexes facilitate future studies on full‐length prion proteins and advanced disease models. ## Plasma Protein Binding Prediction under Scaffold Split: Tree Models, ChemBERTa, Graph Networks, and Late Fusion - Source: ChemRxiv (preprints) - Date: 2026-09-09T00:00:00Z - Categories: Cheminformatics, Property Prediction, ADMET & Safety - Authors: Pradyumna Kumar Pradhan - DOI: 10.26434/chemrxiv.15008530/v1 - External ID: 10.26434/chemrxiv.15008530/v1 - Keywords: SMILES, ChemBERTa, MoleculeNet, Random Forest, XGBoost, graph neural networks, GNNs, ADMET - Source URL: - Dashboard article: Abstract: From idea generation to testing, drug discovery is a lengthy and expensive process. Rather than chasing all of their molecule ideas, scientists prioritize certain molecules that are more likely to be successful, from synthesis to safety to therapeutic use. Machine learning and AI enable researchers to narrow their candidate choices by predicting important properties such as ADMET (Absorption, Distribution, Metabolism, Excretion, and Toxicity). This study uses the Plasma Protein Binding (PPB) dataset from DeepChem’s MoleculeNet, with several graph neural networks (GNNs), ChemBERTa (a language model for SMILES strings of molecules), and fusions of these models. The performance of these neural models is compared against two standard tree-based models: Random Forest (RF) and XGBoost. Instead of random splitting of the dataset, this study evaluates the models based on scaffold splitting that separates molecules by the Bemis–Murcko scaffold and puts the entire group into only the training, validation, or test set. This study demonstrates that for the PPB dataset, GNNs and ChemBERTa models perform better than the tree models. On a mean basis, the fusion models perform better than the individual GNNs or ChemBERTa-only models. Random splitting on the RF model exhibits a better R2 score than on a scaffold-split dataset. Because under scaffold splitting the model predicts on unseen cores, the task becomes more difficult than under regular random splitting. ## ProAlign-DDI: A Zero-Shot Drug–Drug Interaction Event Prediction Framework with Prototype Alignment - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-09T00:00:00+00:00 - Categories: Cheminformatics - Authors: Zhaofeng Niu, Xiaoya Zhao, Ziqi Fan, Yuzhuo Yuan, Defu Qiu, Daohui Ge - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c01621 - Source URL: - Dashboard article: Abstract: The accurate identification of drug–drug interaction (DDI) events (DDIEs) is essential for ensuring medication safety and preventing adverse effects. However, novel drug development constantly generates new DDIE classes that suffer from a severe scarcity of labels, making zero-shot learning an essential solution for emerging cases. Unfortunately, the long-tailed distribution of existing DDI datasets causes overfitting to head classes and poor generalization to tail classes, thereby severely impairing zero-shot performance. To address this limitation, we propose a novel framework, ProAlign-DDI, which integrates a Substructure-Guided Semantic Adaptive Aggregation Module (SAAM) with a Prototype Alignment Module (PAM). In this framework, SAAM adaptively captures task-relevant features to produce high-quality aggregated representations. Subsequently, PAM employs class-level semantic prototypes to regularize these representations. This alignment strategy not only optimizes the global feature space but also alleviates the bias inherent in long-tailed DDI datasets. Comprehensive experiments show that ProAlign-DDI achieves competitive performance in conventional zero-shot learning and more balanced performance in generalized zero-shot learning, confirming its effectiveness in inferring unseen DDIE classes. ## QSAR as a Small-Deformation Theory of Effective Free-Energy Landscapes - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-09T00:00:00+00:00 - Categories: Property Prediction, Targets & Structures - Authors: Jun-ichi Okada, Katsuhito Fujiu - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c01230 - Keywords: QSAR, activity cliffs - Source URL: - Dashboard article: Abstract: Quantitative structure–activity relationships (QSARs) remain widely used in medicinal chemistry, yet their physical interpretation is often limited and the meaning of their applicability domain (AD) is commonly treated in statistical or descriptor-space terms. Here we propose a conceptual framework in which QSAR is interpreted as a small-deformation theory of effective free-energy landscapes. Ligand binding or chemical substitution is represented as a perturbation of an effective free-energy landscape defined over reduced state variables, and the resulting free-energy response is expressed as a contracted observable over the equilibrium ensemble of the reference system via a Zwanzig-type perturbation formula. At the phenomenological level, this response reduces to a low-order expansion in descriptor space, from which conventional linear QSAR emerges as the leading-order approximation. Within the assumed effective description and under stated regularity conditions, the corresponding first-order coefficients are given exactly by the ensemble-averaged sensitivities of the free-energy landscape to descriptor perturbations, without requiring a cumulant truncation. The AD acquires a physical meaning as the region in which a local low-order expansion of the free-energy response remains valid. In the small-deformation regime, this same region further corresponds to ligand-induced landscape perturbations that remain local, smooth, and sufficiently weak over the reference ensemble. Conversely, in systems for which an effective-landscape description is appropriate, QSAR breakdown may be associated with phenomena such as activity cliffs, induced fit, and state-dependent binding, which can involve qualitative reorganization or rapid reweighting of the effective landscape. For a specified thermodynamic system, the second-cumulant term provides a non-positive variance correction to the leading-order ensemble-average free-energy response, arising from the variance of the landscape perturbation over the reference ensemble. This correction reflects a fluctuation-assisted effect in which configurational heterogeneity within the pre-existing ensemble allows Boltzmann reweighting to redistribute statistical weight toward microstates more strongly influenced by the perturbation. Independently, the exact free-energy response is no greater than the leading-order ensemble-average estimate, consistent with Jensen’s inequality. The present framework is conceptual. Its interpretation of QSAR coefficients as ensemble-averaged sensitivities of the free-energy landscape requires additional assumptions and state-resolved information beyond conventional compound-level QSAR data, and it does not by itself provide a prospective AD method. ## Reaction-Condition- and Ensemble-Aware Screening of Dual-Atom ORR Catalysts via Physics-Informed Machine Learning - Source: ACS Catalysis (journals) - Date: 2026-09-09T00:00:00+00:00 - Authors: Prajeet Oza, Victor Fung, Guoxiang Hu - Journal: ACS Catalysis - DOI: 10.1021/acscatal.6c04252 - Keywords: graph neural networks, transformer, equivariant, density functional theory, DFT, ab initio - Source URL: - Dashboard article: Abstract: Metal–nitrogen–carbon dual-atom catalysts (M1/M2–N–C DACs) have emerged as promising alternatives to Pt-based catalysts for the oxygen reduction reaction (ORR), yet their rational discovery is hindered by an enormous chemical and structural design space. Here, we develop a physics-informed machine learning (ML) framework integrated with high-throughput density functional theory (DFT) to systematically screen 22,599 M1/M2–N–C DAC structures spanning 729 transition-metal pairs and 31 structural configurations. Rather than relying on idealized bare-surface models, we construct voltage-dependent ab initio thermodynamic phase diagrams to identify realistic active-site structures and ORR limiting potentials under electrochemical operating conditions. To accelerate screening, we train equivariant transformer graph neural networks on DFT-generated adsorption energetics, achieving high predictive accuracy with mean absolute errors as low as 0.015 eV for adsorption energies and 0.022 V for derived ORR limiting potentials. We further introduce an experimentally relevant descriptor, the percentage of catalytically active structural configurations for each metal pair, which captures the ensemble nature of experimentally synthesized DACs and provides more reliable catalyst ranking than conventional single-configuration approaches. The framework identifies 3431 DAC structures with predicted ORR limiting potentials exceeding 0.8 V and reveals that only a small subset of metal pairs exhibits consistently high activity across configurations. Several top-performing candidates, including Co/Cr, Co/Ag, Co/Ru, Co/Ir, and Co/Zn, are validated by additional DFT calculations and show strong agreement with available experimental trends. In addition, stability screening uncovers multiple DACs predicted to possess both higher ORR activity and greater thermodynamic stability than benchmark Fe/Co–N–C catalysts. This work establishes a scalable ML-assisted paradigm for realistic electrocatalyst screening and provides design principles for the discovery of next-generation dual-atom ORR catalysts. ## Reducing cost of fluorescent microscopy through dye exclusion, distillation techniques, and deep learning - Source: Scientific Reports (journals) - Date: 2026-09-09T00:00:00+00:00 - Authors: Adriana Borowa, Bartosz Zieliński, Dawid Rymarczyk - Journal: Scientific Reports - DOI: 10.1038/s41598-026-69424-3 - Keywords: property prediction - Source URL: - Dashboard article: Abstract: High Content Screening (HCS) via the Cell Painting assay is a powerful drug discovery tool, but the requirement for five fluorescent channels significantly increases the cost and complexity of image acquisition. Current deep learning models for HCS are trained and evaluated on all five channels, and it remains unclear whether robust biological representations can be maintained when channels are removed at inference time. This work introduces Channel-Reduced DINO (CR-DINO), a self-distillation method built on the DINO framework that generates rich HCS image representations from a reduced number of channels. CR-DINO employs a teacher-student architecture in which the teacher retains full five-channel visibility while the student receives progressively fewer channels following a curriculum-inspired schedule. This scheduled reduction, moving from five channels down to two over the course of training, encourages the student to learn cross-channel dependencies rather than relying on information directly available in its input. Experiments on the Bray dataset, validated through Mode of Action prediction, biological activity property prediction, and image–structure retrieval tasks, show that using only two channels (DNA and Mito) provides results on par with those using the full set, and that CR-DINO recovers performance even for the least informative channel pair (ER and AGP). These results highlight the potential of our method to significantly reduce the reagent and acquisition costs of consecutive HCS experiments without sacrificing the depth of biological insights. ## Reliable Machine Learning for DPPH Antioxidant QSAR: Scaffold- and Study-Aware Validation with an Out-of-Domain Sulfonamide-Tyr-Gly Case Study - Source: ChemRxiv (preprints) - Date: 2026-09-09T00:00:00Z - Categories: Cheminformatics, Property Prediction - Authors: Dennis Obinna Orji, Ike Ozoemena Christian - DOI: 10.26434/chemrxiv.15008549/v1 - External ID: 10.26434/chemrxiv.15008549/v1 - Keywords: RDKit, QSAR, Random Forest, pIC50 - Source URL: - Dashboard article: Abstract: Machine-learning QSAR models can prioritize antioxidant compounds, but apparent accuracy depends strongly on how chemical data are partitioned for validation. This study develops a reproducible DPPH antioxidant QSAR workflow using 1,891 curated molecules from a public 30 min DPPH dataset. Molecular structures were represented by nine RDKit physicochemical descriptors and 2,048-bit Morgan fingerprints (radius 2), giving 2,057 combined features.Random Forest regression was evaluated as the primary model,with Extra Trees as an independent ensemble comparator.Random Forest achieved mean R-squared = 0.712 and MAE = 0.297 pIC50 units under random five-fold cross-validation, but performance decreased to R-squared = 0.447 and MAE = 0.433 under Bemis-Murcko scaffold-aware validation, and to R-squared = 0.124 and MAE = 0.545 under DOI-grouped validation. In an internal held-out set, molecules with scaffolds represented during training performed better than those with unseen scaffolds(R-squared=0.824 versus 0.580).Nearest-neighbour Morgan Tanimoto analysis further separated structurally supported predictions from extrapolation: inside-domain molecules achieved R-squared = 0.830, whereas outside-domain performance was R- squared =-0.031. The workflow was then applied to two experimentally investigated sulfonamide-Tyr-Gly derivatives. Both had low maximum Tanimoto similarity(0.364 and 0.333), unseen scaffolds,and were overpredicted by approximately one pIC50 unit.The results show that random-split performance alone is insufficient for judging DPPH-QSAR reliability and that chemically informed validation and applicability-domain analysis are necessary when predictions are extended to structurally novel compounds. ## Synthesis and in silico studies of hexahydroacridine derivatives as potential antitumor agents - Source: Scientific Reports (journals) - Date: 2026-09-09T00:00:00+00:00 - Categories: Docking & Screening, Free Energy & MD - Authors: Eman A. E. El-Helw, Sherif F. Hammad, Hassan A. Khatab, Youssef A. Said, Esmail M. El‐Fakharany, Ahmed I. Hashem - Journal: Scientific Reports - DOI: 10.1038/s41598-026-68544-0 - Keywords: Molecular docking, kinase, Molecular Dynamics, metabolic stability - Source URL: - Dashboard article: Abstract: Despite advances in targeted cancer therapies, drug resistance and off-target toxicity remain major challenges, underscoring the need for new multi-targeted agents with improved selectivity. The present study evaluated the anticancer potential of newly synthesized hexahydro-1,8-acridinedione derivatives featuring strategic structural modifications at positions 9 and 10. The novelty of this work lies in the introduction of N -aryl substituents combined with C-9 biphenyl or 4-bromophenyl groups, a structural motif not previously reported for this chemotype. This dual modification was rationally designed to enhance π-π stacking interactions with kinase ATP-binding pockets while improving metabolic stability through fluorine substitution. Compounds were synthesized via a one-pot multicomponent reaction catalyzed by PTSA and characterized by IR, 1 H/ 13 C NMR, and LC–MS. Guided by structure-based design principles, five derivatives ( 4a, 4d, 4f., 5a, 5b ) were evaluated for cytotoxicity against HSF, H460, A431, A549, and MDA-MB-231 cell lines. Compound 4d emerged as the most potent, with an IC₅₀ of 17.91 ± 2.8 µg/mL against H460 lung cancer cells and a high selectivity index (SI = 20.7), in addition to exhibiting a considerable anti-telomerase activity. Molecular docking revealed strong binding affinities of 4d with TOP2B (− 7.02 kcal/mol), p38 MAPK (− 7.05 kcal/mol), p53 (− 7.25 kcal/mol), and EGFR (− 6.55 kcal/mol), supported by MM-GBSA binding free energies (ΔG bind ≈ − 42 kcal/mol for 4d–EGFR). Furthermore, 100-ns Molecular Dynamics simulations confirmed the stability of the 4d -EGFR complex (RMSD ≈ 1.31 Å; persistent H-bonds with CYS773/GLN767). These findings establish compound 4d as a promising multi-target lead scaffold, warranting further optimization toward selective anticancer therapeutics. ## Synthesis of Sb(iii)-immobilized on nitrogen-doped mesoporous silica nanotubes for efficient and green preparation of pyrazolopyranopyrimidines: comprehensive characterization, green chemistry evaluation and computational antifungal profiling against 5TZ1 - Source: RSC Advances (journals) - Date: 2026-09-09T00:00:00Z - Categories: Docking & Screening - Journal: RSC Advances - DOI: 10.1039/d6ra06424h - External ID: 486fee4c2de95154ee90a358213f602fe8de817f - Keywords: molecular docking - Source URL: - Dashboard article: Abstract: In this study, a novel heterogeneous nanocatalyst, Sb(iii) immobilized on nitrogen-doped mesoporous silica nanotubes (N-MSNTs/Sb(iii)), was successfully synthesized. This catalyst was then applied in a one-pot, four-component synthesis of pyrazolopyranopyrimidine derivatives via the condensation of ethyl acetoacetate, hydrazine hydrate, aromatic aldehydes, and barbituric/thiobarbituric acid. Comprehensive characterization using FT-IR, FE-SEM, EDS, HR-TEM, XRD, and BET/BJH analyses confirmed its structural features. Notably, HR-TEM revealed hollow nanotubular structures with an inner diameter of ∼23 nm and a wall thickness of 27 nm. Moreover, BET/BJH measurements exhibited a type IV isotherm, indicative of a highly porous structure, with a remarkable specific surface area of 1255 m2 g−1, a pore volume of 0.71 cm3 g−1, and an average pore diameter of 2.26 nm. Catalytic evaluations demonstrated the superior performance of N-MSNTs/Sb(iii) in aqueous media at room temperature, affording the target compounds in high yields (78–96%) within short reaction times (35–85 min). This remarkable efficiency stems from synergistic activation of support by the nitrogen sites and active Sb(iii) Lewis acidic species. Furthermore, the catalyst exhibited excellent stability, maintaining its initial activity over five consecutive reuse cycles. Finally, molecular docking simulations were performed against the antifungal target 5TZ1. The results confirmed that the carbonyl and NH groups within the synthesized molecular scaffolds establish favorable binding interactions, highlighting the promising drug-like properties of the final products. ## Synthesis, Pharmacological Evaluation, Molecular Docking, and In Silico ADMET Profiling of Novel Ibuprofen–Chalcone Hybrids as Potent Anti-Inflammatory Agents - Source: Russian Journal of Bioorganic Chemistry (journals) - Date: 2026-09-09T00:00:00Z - Categories: Docking & Screening, ADMET & Safety - Journal: Russian Journal of Bioorganic Chemistry - DOI: 10.1134/S1068162025603337 - External ID: 0a27f307ff7b583cb084e9c6982a23b1a79b1800 - Keywords: Molecular Docking, ADMET - Source URL: - Dashboard article: - Abstract: not stored for this record. ## The Epoxybenzooxocine Core of Integrastatins and Analogs: A Comprehensive Review of Structure, Synthetic Approaches, and Biological Applications - Source: ACS Omega (journals) - Date: 2026-09-09T00:00:00+00:00 - Categories: Cheminformatics, Docking & Screening, ADMET & Safety - Authors: Aida S. Rakhimzhanova, Irina A. Pustolaikina, Alfiya F. Kurmanova, Ruslan A. Muzaparov, Darya D. Kapishnikova, Alena L. Stalinskaya, Ivan V. Kulakov - Journal: ACS Omega - DOI: 10.1021/acsomega.6c06756 - Keywords: pharmacophore mapping, Lipinski, drug likeness, PubChem, bioactivity, carcinogenicity, ADMET - Source URL: - Dashboard article: Abstract: Integrastatins and their biogenetic counterparts possess a rare heterotetracyclic framework containing an unprecedented \[6/6/6/6\]-tetracyclic skeleton and a central \[3/3/1\]-bicyclic ketal core with high pharmaceutical potential, particularly due to their potent anti-HIV-1 integrase activity. Despite their promising scaffolds, a systematic consolidation of their chemistry and biology has been entirely lacking, creating a critical gap between the complex architecture of these natural polyketides and the targeted design of drug-like inhibitors. Herein, we present the first comprehensive review of this field, systematically analyzing 42 peer-reviewed publications from 1999 to 2025 retrieved from the Scopus, Web of Science, PubMed, and PubChem databases. Structural analysis confirms that while these rigid, nonplanar V-shaped molecules comply with Lipinski’s Rule of Five as a baseline filter for drug-likeness, deeper ADMET profiling further highlights their high predicted human intestinal absorption (HIA) alongside a low risk of cardiotoxicity. Synthetically, we detail how the field has fundamentally evolved from an arduous 11-step sequence in 2003 to highly streamlined, atom-economical one-step protocols in 2008 and 2021, culminating in the 2025 total syntheses that successfully resolved long-standing structural misinterpretations of epicoccolide A and epicocconigrone A. Furthermore, a comparative assessment across eight critical synthetic parameters highlights the operational trade-offs between step efficiency, scalability, and functional group tolerance in current cascade cyclizations. Biologically, we evaluate the extensive in vitro and in silico profiling of over 30 synthetic derivatives, highlighting their antimicrobial (M. bovis, S. aureus, P. carotovorum, C. albicans), antitubercular, and antiviral (SARS-CoV-2 and HIV-1 integrase inhibition) potencies. Integrating pharmacophore mapping with in silico ADMET profiling demonstrates that despite favorable absorption and cardiac safety, predicted liabilities such as human hepatotoxicity, carcinogenicity, and skin sensitization highlight critical areas for targeted structural modification. Ultimately, by bridging synthetic methodology with in silico and experimental bioactivity evaluation, this review establishes epoxybenzooxocine-based heterocycles as viable early-stage lead platforms and outlines strategic directions to advance these scaffolds in targeted drug discovery. ## Torsionator: Dihedral Parameterization and Conformer Search Made Easy - Source: Journal of Chemical Theory and Computation (journals) - Date: 2026-09-09T00:00:00+00:00 - Categories: ML Potentials - Authors: Giorgia Brosio, Alice Triveri, Pietro Vidossich, Sergio Decherchi, Marco De Vivo - Journal: Journal of Chemical Theory and Computation - DOI: 10.1021/acs.jctc.6c01006 - Keywords: Neural network potentials, MACE - Source URL: - Dashboard article: Abstract: Neural network potentials (NNPs), trained on quantum mechanical (QM) data, can deliver near QM-level accuracy while being much faster than QM calculations. In this work, we introduce a new software tool, Torsionator, which currently supports MACE, OBIWAN, and UMA NNPs for energy minimization of small molecules. Torsionator can analyze a set of selected conformers to parameterize the dihedral angles. These dihedral parameters can then be used for molecular mechanics calculations and simulations. We showcase Torsionator by performing dihedral scanning and parameterization of two representative antiviral compounds: favipiravir defluoro analog T1105 and emivirine. To thoroughly evaluate its robustness and general applicability, we benchmarked Torsionator using MACE on the TorsionNet500 dataset, comprising 500 chemically diverse small molecules with reference QM torsional profiles. These results indicate Torsionator as a practical, efficient, and scalable software tool for dihedral parameterization, enabling its routine integration into molecular simulation workflows. ## UPLC-MS metabolite profiling and antioxidant efficacy of Neopestalotiopsisclavispora isolated from Oroxylum indicum (L.) Kurz Stem Bark - Source: PLOS One (journals) - Date: 2026-09-09T00:00:00Z - Categories: Docking & Screening - Journal: PLOS One - DOI: 10.1371/journal.pone.0357595 - External ID: 6974ed4b9e548b5392f8199b6aa96c78e86a6d84 - Keywords: Molecular docking, enzyme, Molecular dynamics - Source URL: - Dashboard article: Abstract: Endophytic fungi represent a valuable source of bioactive compounds with diverse pharmacological properties. In this study, we explored the antioxidant potential of secondary metabolites produced by Neopestalotiopsis clavispora (N.clavispora), an endophytic fungus isolated from Oroxylum indicum Kurz stem bark. N.clavispora extract was subjected to UPLC-MS analysis to profile its bioactive components, and its antioxidant capacity was assessed using the DPPH (1,1-diphenyl-2-picrylhydrazyl) free radical scavenging assay. For in vivo evaluation, Swiss albino mice were administered with 100 mg/kg bodyweight dose of extract for five consecutive days, followed by exposure to 2 Gy of gamma radiation at a dose rate of 3.58 Gy/min. Subsequent measurements of superoxide dismutase (SOD) activity in liver homogenate were conducted to assess antioxidant enzyme response. The data obtained from UPLC-MS, showed, cimicifoetiside B (identified by UNIFI database matching; 44% relative abundance) as the predominant metabolite. N. clavispora extract showed moderate DPPH scavenging (IC₅₀ 82.86 μg/mL vs ascorbic acid 19.01 μg/mL). Molecular docking using cimicifoetiside B against the antioxidant enzyme Superoxide dismutase (SOD), revealing a docking score of −4.96 kcal/mol with 1 hydrogen bond. Molecular dynamics simulations confirmed the interaction with a binding energy of –12.36 ± 02.32 kcal/mol. Binding affinities and molecular interaction profiles indicated possible association with SOD. In vivo, N.clavispora extract at 100 mg/kg body weight/day yielded SOD activity of 1.98 ± 0.38 vs. radiation group 0.78 ± 0.13. These results suggest N.clavispora may have antioxidant related activity under the conditions tested, supporting further in vitro and in vivo studies. ## What actually moves virtual screening performance: two measured resolution limits and six pre-registered interventions - Source: ChemRxiv (preprints) - Date: 2026-09-09T00:00:00Z - Categories: Cheminformatics, Property Prediction, Docking & Screening, Targets & Structures - Authors: John Goodman - DOI: 10.26434/chemrxiv.15008496/v2 - External ID: 10.26434/chemrxiv.15008496/v2 - Keywords: DiffDock, Boltz 2, SMILES, AutoDock Vina, virtual screening, receptor - Source URL: - Dashboard article: Abstract: Reported improvements to structure-based virtual screening are usually differences in AUROC or enrichment between two protocols on one benchmark. We asked what a benchmark of typical size can actually resolve, and then measured six interventions against that limit with the reading rule for each committed to version control before the data existed. Two limits, not one. Scoring the same Vina poses with a second implementation of the same scoring function — agreeing at r = 0.970, no compound differing by more than 2 kcal/mol — moves AUROC by 0.020 \[95% CI 0.011, 0.029\]. Re-running the whole protocol at a different search effort and scoring it the same way moves AUROC by 0.039 \[0.024, 0.056\]. The first bounds a comparison of scoring functions on fixed poses; the second bounds a comparison of protocols, which is what most published improvements are. Both fall inside the range in which docking improvements are typically reported. On SARS-CoV-2 Mpro: changing the scoring function on identical poses moved AUROC +0.140, repairing a receptor preparation defect +0.045, scoring the pose ensemble rather than the top pose +0.055, supplying the correct binding site to a co-folding model −0.013, and eight-fold search effort −0.019. Two clear the relevant floor, one is marginal and two fall inside. The sixth — substituting DiffDock-L pose generation — is not evaluable on this panel: DiffDock places a fifth of the compounds outside the binding site, and that fifth is heavier and more often active than the panel, so any comparison on the remainder is biased. These are distributions over cross-validation fold assignment, not point estimates: we previously published the pose-ensemble arm as +0.019 and refuted, and it proved to be the 2nd percentile of a distribution centred at +0.031. The search-effort result is the sharpest either way: its measured effect is smaller than the run-to-run reproducibility of the protocol it was measured in. Docking adds nothing demonstrable over seven free physicochemical descriptors on either panel we assembled: on Mpro it significantly subtracts (−0.002 \[−0.004, −0.000\]) and on Factor Xa its +0.015 \[−0.003, +0.034\] crosses zero. Applying the same descriptors to LIT-PCBA gives a median AUROC of 0.7260 across 15 targets against published AutoDock Vina at 0.581 and GNINA at 0.611–0.616. These descriptor baselines are supervised on each benchmark's own labels while docking is not, so this is a bias control in the sense of Sieg et al., not a claim that descriptors are a deployable substitute. On the AVE-debiased variant the descriptor median falls to 0.6510. One method exceeded the floor. Boltz-2, an open-weights co-folding model scoring compounds from sequence and SMILES alone, does not beat the descriptor baseline on its own (Mpro +0.026, Factor Xa +0.019, both intervals spanning zero) but adds +0.093 \[+0.062, +0.125\] and +0.075 \[+0.051, +0.099\] respectively when combined with those descriptors. Its advantage is not that it is less property-driven — descriptors explain 63% of its score and 63% of Vina's — but that on Mpro its residual, the part orthogonal to those descriptors, ranks actives at 0.657 where six docking scores residualised the same way sat between 0.494 and 0.573. Both descriptor comparisons are supervised on each target's own labels; transferred cold to the other target the descriptor model ranks actives below chance , and the standalone Boltz-2 score — which requires no labels at all — is the only configuration usable on a target with no known actives. Across twelve cross-validation fold seeds the two do not overlap on Mpro: the highest docking residual observed is 0.581, the lowest Boltz-2 residual 0.644. On Factor Xa the residual result replicates in direction — 0.5882 \[0.5789, 0.5975\] against that target's own Vina arm at 0.5733 — but the margin is +0.015 rather than +0.083 and the ranges touch at their extremes, so the size of the effect does not replicate. One point outside a band is not a generalisation, so we pre-registered a third scoring approach to test it. FlashBind, an EGNN affinity head over learned docking poses, residualises to 0.538 \[0.529, 0.546\] — inside the docking band — with all three pre-registered controls passing and a pocket-agreement check ruling out the confound that the band is a property of our docking box. The claim that this ceiling belongs to the scoring approach rather than to these panels therefore remains untested rather than supported. ## WikiPathways: biological pathways across the species - Source: chem-bla-ics (Linked Chemistry) (feeds) - Date: 2026-09-09T00:00:00+00:00 - Categories: Blog - Source URL: - Dashboard article: - Abstract: not stored for this record. ## “Green spectrophotometric determination of rupatadine, its toxic impurity, and methyl paraben: comparative AI-assisted univariate versus multivariate chemometrics” - Source: Scientific Reports (journals) - Date: 2026-09-09T00:00:00+00:00 - Categories: Spectra & Analytical - Authors: Salma S. Mansour, Amr M. Mahmoud, Azza A. Moustafa, Nancy W. Nashat - Journal: Scientific Reports - DOI: 10.1038/s41598-026-68686-1 - Keywords: s, chemometric - Source URL: - Dashboard article: Abstract: Sensitive and selective green UV-spectrophotometric univariate and multivariate chemometric methods were developed to resolve the overlapped spectra of rupatadine fumarate (RUP), its impurity desloratadine (DES), and methyl paraben (MP). The univariate approach employed was the dual-wavelength (DW) method. Guided by artificial intelligence (AI) in the selection of the optimal wavelength pair, the determination of RUP was performed at 230 and 284.3 nm, and for MP at 236.6 and 253.8 nm. Among the multivariate methods, ANN demonstrated superior predictive accuracy (lowest RMSEP values), followed by PLS-1 and MCR-ALS, highlighting the power of machine learning for complex mixture resolution, particularly in the presence of the preservative MP. A three-factor, five-level experimental design was adopted to construct a calibration set comprising 25 mixtures with varying component ratios, along with 6 independent validation mixtures. The developed methods were successfully applied to pharmaceutical formulation containing the studied drug and were validated in accordance with International Conference of Harmonization (ICH) guidelines. The results obtained were reliable and reproducible, confirming the suitability of these methods for routine analysis and quality control in laboratories. Moreover, the GLANCE tool (Graphical Layout for Analytical Chemistry Evaluation) offered a clear visual summary of the twelve structured method attributes. ## Quantum-accurate atomistic modeling of enzyme catalysis using a machine learned potential - Source: arXiv (preprints) - Date: 2026-09-08T18:00:07Z - Categories: ML Potentials - Authors: Meng Gao, Armin Shayesteh Zadeh, Aniruddha Seal, Siva Dasetty, Siddarth K. Achar, Misko Dzamba, Benjamin K. Miller, Leif D. Jacobson, C. Lawrence Zitnick, Brandon M. Wood, Zachary W. Ulissi, Daniel S. Levine, Andrew L. Ferguson - External ID: 2609.09293v1 - Keywords: MLIP, enzyme, kinase - Source URL: - Dashboard article: - PDF: - Code: Abstract: Electronic rearrangements associated with bond forming/breaking in catalytic enzymes require quantum mechanical (QM) treatment beyond classical molecular mechanics (MM). Hybrid QM/MM methods enable tractable simulations but require system-specific setup and are sensitive to the QM region choice and treatment of the QM/MM interface. We demonstrate quantum-accurate treatment of all-atom, complete enzymes in explicit solvent comprising up to 54k atoms and 1 microsecond of total simulation time using the machine-learned interatomic potential (MLIP) eSEN-omol. We reproduce experimental barrier trends for Claisen rearrangement in chorismate mutase, resolve critical intermediate states in PETase catalyzed polymer depolymerization, and distinguish mechanistic alternatives for metal-activated phosphoryl transfer in nucleoside diphosphate kinase. We realize 1000x speedups relative to typical QM/MM calculations without system-specific tuning. These results establish MLIPs as a practical route to QM-accurate simulations of enzyme catalysis. ## Accessible hybrid DFT-quality NMR crystallography via gas-phase Machine Learning Interatomic Potentials - Source: Chemical Science (journals) - Date: 2026-09-08T13:03:07Z - Authors: Shubha Gunaga, Rob Schurko, Sean T. Holmes, Frederic Mentink-Vigier - Journal: Chemical Science - DOI: 10.1039/d6sc04941a - Keywords: DFT, density functional theory - Source URL: - Dashboard article: Abstract: Nuclear magnetic resonance (NMR) crystallography is a robust method for structure determination, but its reliance on density functional theory (DFT) for geometry refinement limits its speed and accessibility. Recent machine‑learning... ## Multi-ligand simultaneous docking of Carica papaya leaf phytochemicals, Carpaine and Rutin, reveals multi-mechanism inhibition of cancer proteins BCL-2 and WWP1 - Source: arXiv (preprints) - Date: 2026-09-08T10:29:10Z - Categories: Docking & Screening, ADMET & Safety - Authors: Merla Sudha, Asmita Saha, Belaguppa Manjunath Ashwin Desai, Anil Ranu Mhashal, Pronama Biswas - DOI: 10.1016/j.phyplu.2025.100829 - External ID: 2609.08547v1 - Keywords: molecular docking, binding affinity, molecular dynamics - Source URL: - Dashboard article: - PDF: Abstract: Cancer remains a major global health concern due to chemotherapy resistance and toxicity from high-dose treatments. To overcome these challenges, new therapeutic strategies targeting key proteins in cancer progression are essential. This study evaluates two phytochemicals, Carpaine (Car) and Rutin (Rut), from Carica papaya leaves, for their potential in enhancing cancer therapy by targeting B-cell lymphoma 2 (BCL-2) and WW domain-containing protein 1 (WWP1) proteins. We assessed their additive, allosteric, and synergistic effects using molecular docking, multi-ligand simultaneous docking (MLSD), molecular dynamics (MD) simulations, and MMPBSA analysis. Car and Rut showed an additive effect on BCL-2 by binding at distinct regions within the same pocket. MLSD revealed an improved binding affinity of -13.13 +/- 0.08 kcal/mol, compared with individual ligands or the commercial inhibitor Venetoclax. For WWP1, Car bound near the H-site and Rut near the Le-site, exhibiting an allosteric effect that increased Car's binding affinity in MLSD to -15.59 +/- 0.39 kcal/mol. Furthermore, Rut combined with bortezomib (Bort) demonstrated a synergistic interaction with WWP1. Binding energies were -7.64 +/- 0.156 kcal/mol for Bort, -10.26 +/- 0.07 kcal/mol for Rut, and -15.59 +/- 0.39 kcal/mol for MLSD, suggesting a more stable complex through synergy. These results suggest Car and Rut, particularly in combination with Bort, as promising candidates against cancer-related proteins BCL-2 and WWP1. Further experimental validation is warranted to explore their therapeutic potential. ## TSBench: A physics-grounded benchmark for evaluating LLM understanding of chemical reaction mechanisms - Source: arXiv (preprints) - Date: 2026-09-08T09:44:37Z - Categories: Reaction Informatics, LLMs & Agents - Authors: Xiaohu Xu, Tong Zhu - External ID: 2609.08503v1 - Keywords: LLM, LLMs, synthesis planning - Source URL: - Dashboard article: - PDF: Abstract: Understanding a chemical reaction requires mapping a symbolic reactant-product description onto the three-dimensional pathway through which atoms rearrange, yet chemistry benchmarks for large language models (LLMs) largely probe factual knowledge and text-based reasoning. Here we introduce TSBench, a benchmark in which an LLM agent uses structure-editing tools to construct three-dimensional transition-state (TS) guesses verified by an automated quantum-chemical pipeline, yielding a physics-grounded pass/fail verdict. Across 546 evaluations of seven frontier LLMs on 78 elementary reactions, the aggregate success rate rose from 50.4% to 66.8% under diagnosis-driven revision; the best models approached 90% on the simplest reactions, yet performance dropped sharply with mechanistic complexity. Most failed attempts produced locally plausible saddle points whose reaction paths led to the wrong reactant-product pair, revealing local geometric intuition without a robust grasp of the global reaction coordinate. TSBench establishes a mechanism-level yardstick for LLM agents in mechanism-sensitive tasks such as synthesis planning and autonomous experimentation. ## An Interpretable Framework Applying Protein Words to Predict Protein–Small Molecule Complementary Pairing Rules - Source: Chemical Science (journals) - Date: 2026-09-08T09:23:01Z - Authors: Jingke Chen, Jingrui Zhong, Tazneen Hossain Tani, Zidong Su, Xiaochun Zhang, Boxue Tian - Journal: Chemical Science - DOI: 10.1039/d6sc03365b - Source URL: - Dashboard article: Abstract: Despite the high accuracy of 'black box' deep learning models, drug discovery still relies on protein-ligand interaction principles and heuristics. To improve interpretability of protein-small molecule binding predictions, we... ## Predicting directional flexibility in proteins - Source: arXiv (preprints) - Date: 2026-09-08T09:18:46Z - Categories: Property Prediction, Design de novo, Free Energy & MD - Authors: Vsevolod Viliuga, Leif Seute, Matteo Tadiello, Nicolas Wolf, Frauke Gräter, Arne Elofsson - External ID: 2609.08474v1 - Keywords: graph neural network, generative models, equivariant, Molecular Dynamics - Source URL: - Dashboard article: - PDF: - Code: Abstract: Predicting protein dynamics is a long-standing problem in computational structural biology. Often, protein function critically depends on local directed motions, such as hinge movements, catalytic loop rearrangements and domain reorientations, which can be characterized by directional flexibility and correlated structural motions of the protein backbone. While Molecular Dynamics (MD) simulations provide an established but often prohibitively expensive approach, recent deep generative models aim to reduce this cost by directly predicting conformational ensembles, emulating MD. However, due to their large size and the need to generate several states until the derived dynamical properties converge, these models remain expensive. In this work, we propose BackFlip-2: a fast SE(3)-equivariant graph neural network trained to directly predict dynamical descriptors, such as directional backbone flexibility and pairwise dynamic correlations, from an equilibrium structure. In a series of experiments, we show that our model matches the accuracy of substantially larger ensemble generation models while being orders of magnitude faster, and demonstrate that the proposed equivariant architecture is especially well-suited for capturing anisotropic motions in proteins. BackFlip-2 model weights, training and inference code are available at https://github.com/graeter-group/backflip. ## From London to Morse via Binnig, Quate, and Gerber - Source: arXiv (preprints) - Date: 2026-09-08T08:32:50Z - Authors: Sofia Alonso Perez, Matthew O. Blunt, Frederick Carlisle, Neil R. Champness, Janette L. Dunn, Matthew Edmondson, Rowan Evers, Connor Fields, Subhashis Gangopadhyay, James Hayton, Samuel P. Jarvis, Filipe Junqueira, Lev Kantorovich, Brian Kiraly, Natalio Krasnogor, Ioannis Lekkas, Morten Møller, Philip Moriarty, Chris Pakes, Emmanuelle Pauliac-Vaujour, Oliver Phillips, Adrian Radocea, Philipp Rahe, Mohammad Abdur Rashid, Hongqian Sang, Alex Saywell, Nikhil Seeja Sivakumar, Peter Sharp, Andrew Stannard, Julian Stirling, Adam Sweetman, Simon Taylor, Richard A. J. Woolley - External ID: 2609.08425v1 - Source URL: - Dashboard article: - PDF: Abstract: In their landmark paper introducing the atomic force microscope \[Phys. Rev. Lett. \\textbf\{56\}, 930 (1986)\], Binnig, Quate, and Gerber presciently anticipated that the technique would ultimately be capable of probing interactions running the gamut from weak van der Waals interactions to strong covalent bonding. They also highlighted that the tip-sample forces central to AFM are present, and often highly influential, in scanning tunnelling microscopy; indeed, this realisation directly inspired the invention of the force microscope. In this perspective for the \\textit\{Forty Years of AFM\} special issue, we review selected aspects of two decades of work from our group at the University of Nottingham that span the force range highlighted by BQG and are united by a common, central theme: the probe as active participant rather than passive observer. Our selection of results also covers length- and correlation-scales from the microscopic right down to the single chemical bond limit, tracking a spectrum of interactions from van der Waals/Hamaker forces, through hydrogen bonding, to covalent bonds and, finally, atom-by-atom assembly of metal clusters via vertical tip-sample transfer. Echoing BQG's own observations on the prevalence of probe-sample forces in STM, we also discuss recent evidence that tip-induced heterogeneity underpins first-passage dynamics in molecular diffusion and highlight the challenges in acquiring non-invasive measurements of diffusion barriers for adsorbed molecules that are readily perturbed by the probe. We close with a perspective on machine learning's growing role in automating tip-driven atomic and molecular manipulation. ## MLIP Detective: Active Failure Mode Discovery Beyond Benchmark Scores for Machine-Learning Interatomic Potentials - Source: arXiv (preprints) - Date: 2026-09-08T08:07:33Z - Categories: LLMs & Agents, ML Potentials - Authors: Ryuhei Okuno, Nontawat Charoenphakdee, Kaoru Hisama, Yuta Tsuboi - External ID: 2609.08399v1 - Keywords: MLIP, MACE - Source URL: - Dashboard article: - PDF: Abstract: Universal machine-learning interatomic potentials (u-MLIPs) aim to generalize across diverse configurations. Benchmarks enable reproducible evaluation but may not expose failures outside their predefined scope. Here, we show that physics-informed search can complement benchmark-based evaluation by uncovering hidden failure modes. We introduce MLIP Detective, an agentic framework for active failure mode discovery. Starting from benchmark evidence, MLIP Detective generates falsifiable, physics-informed failure hypotheses, screens them with inexpensive simulations, and escalates only the most suspicious cases to human experts together with proposed verification protocols. Without issue-specific prompting, MLIP Detective identified and characterized a systematic anomaly in MACE-MPA-0: the model predicted some relaxed adsorbate-surface systems involving O- or F-containing adsorbates to be higher in energy than their corresponding separated fragments. Using cross-model comparisons, MLIP Detective further inferred a likely training-data origin for the anomaly, consistent with recent reports. ## Fixed-Dimensional Latent Flow for Generating Variable-Size 3D Molecules - Source: arXiv (preprints) - Date: 2026-09-08T07:03:51Z - Authors: Weichi Yao, Cameron Gruich, Bryan R. Goldsmith, Yixin Wang - External ID: 2609.08333v1 - Keywords: Transformer, Transformers, Equivariant, density functional theory, DFT - Source URL: - Dashboard article: - PDF: Abstract: In molecular discovery, molecule size is coupled to composition, structure, and other target properties. Yet most 3D generators require molecule size to be specified before generation. Here, we introduce Equivariant-Free Transformer-Autoencoded Latent Flow Matching, a two-stage generative framework that relies entirely on a single fixed-dimensional molecule-level latent representation to generate variable-size molecules. The second-stage flow matching model samples this latent vector, and an autoregressive Transformer decoder then determines molecule size while generating atom types, coordinates, and chemically informative states. Canonical atom ordering and rigid-pose alignment enable standard Transformers without equivariant layers, while joint decoding of molecular geometry and an enriched chemical state enables reliable, deterministic, chemistry-guided graph recovery without requiring a learned dense pairwise bond decoder. The same fixed-dimensional latent supports unconditional and property-conditioned flow matching, while optional property supervision adds an internal ranking readout, with no separate predictor or reference calculations. On PCQM4Mv2, EF-TALFM achieves the highest fraction of molecules that are unique, training-set novel, pass sanitization and PoseBusters sanity checks, 89.4\\%, compared with 75.6\\% for UAE-3D and 69.8\\% for FlowMol. EF-TALFM also achieves higher measured computational throughput for training and sampling. Across ten target HOMO--LUMO gaps, internal ranking doubles the density functional theory (DFT)-verified hit rate within $0.1\\,\\mathrm\{eV\}$, while preserving 97\\% novelty among unique verified hits. These results demonstrate that fixed-dimensional molecule-level generation followed by symmetry-resolved autoregressive realization provides a practical architecture for open-ended and property-directed 3D molecular design. ## SMILES2Docking: An Open‐Source Desktop Workflow for Ligand Ionization, Stereochemistry‐Aware Preparation and Semi‐Empirical 3D Refinement - Source: Molecular Informatics (journals) - Date: 2026-09-08T05:44:37Z - Categories: Cheminformatics - Authors: Daniel Andrés Grajales Ruiz, Bruna Flôres Negrisoli, Isabel Cristina Conceição Periquito, Nailton Monteiro do Nascimento‐Júnior, Adriano Marques Gonçalves - Journal: Molecular Informatics - DOI: 10.1002/minf.70053 - Keywords: SMILES, RDKit, graph neural network - Source URL: - Dashboard article: - Code: Abstract: SMILES2Docking converts spreadsheet‐based SMILES libraries into docking‐ready 3D ligands through a single, configurable pipeline. Salt and coformer removal, a three‐mode stereocenter policy, prediction of the dominant protonation state with a graph neural network pKa model (MolGpKa) refined by iterative titration, distance‐geometry embedding with RDKit, a molecular‐mechanics optimization cascade, and an optional semi‐empirical refinement under an implicit solvation model are combined in one tool, with records processed either sequentially or distributed across CPU cores. Users select among permissive, strict, and enumerative stereocenter modes, so the same tool serves broad exploratory screening and precision docking. A per‐compound JSON audit report records the ionization and stereocenter decisions taken for every structure. The tool is distributed as a Python package, a Windows executable with bundled MOPAC, a Linux portable bundle, and a macOS bundle, under GPL‐2.0‐or‐later, and is freely available to non‐commercial users at https://github.com/amgoncalvesusp/Smiles2Docking (DOI: 10.5281/zenodo.20617898 ). ## PocketVE: Stable and Property-Guided Structure-Based Drug Design with Variance-Exploding Diffusion - Source: arXiv (preprints) - Date: 2026-09-08T01:19:17Z - Categories: Property Prediction, Design de novo - Authors: Peining Zhang, Jinbo Bi - External ID: 2609.08101v1 - Keywords: molecule generation, denoising - Source URL: - Dashboard article: - PDF: Abstract: Protein-conditioned 3D molecule generation is a central challenge in structure-based drug design, requiring a balance between pocket compatibility, molecular properties, and physical geometry. We propose \\textbf\{PocketVE\}, a protein-pocket-conditioned variance-exploding (VE) diffusion framework that couples stable coordinate denoising with inference-time property guidance. Specifically, PocketVE combines an EDM-style training and sampling setup for 3D denoising, classifier-free guidance for multi-property steering without external property classifiers, and adaptive protein perturbation as a training-time pocket regularizer. Evaluated on CrossDocked2020 under the GenBench3D protocol, PocketVE improves Valid$\_\{3\\text\{D\}\}$ from 58.6 to 80.6 and reduces strain energy from 457.4 to 127.9 relative to its TAGMol architectural baseline, while retaining competitive docking and molecular-property scores under moderate guidance. A guidance-scale study shows that moderate guidance gives a favorable balance between target-related objectives and geometric quality, whereas stronger guidance can degrade geometry and distributional fidelity. Pocket-permutation and PoseCheck diagnostics further support pocket-specific spatial compatibility with reduced steric conflicts. Overall, the results suggest that geometric stability and inference-time property guidance should be considered as coupled design objectives. ## A Bacillus amyloliquefaciens serine protease reduces ELISA-detectable epitopes of major indoor allergens: Computational modeling of the Fel d 1 binding interface. - Source: The Science of the total environment (journals) - Date: 2026-09-08T00:00:00Z - Categories: Docking & Screening - Journal: The Science of the total environment - DOI: 10.1016/j.scitotenv.2026.182298 - External ID: 52edf76812323293cac091b3378b9d26cdcadd1e - Keywords: molecular docking, enzyme, molecular dynamics - Source URL: - Dashboard article: Abstract: Indoor allergens from pets and dust mites are persistent biological pollutants that pose significant risks to human health, yet effective and environmentally sustainable control strategies remain limited. Here, we isolated Bacillus amyloliquefaciens Ba2501, which secretes a subtilisin-family serine protease. Whole-cell co-culture assays demonstrated that Ba2501 reduces ELISA-detectable epitopes of three major indoor allergens-Fel d 1 (cat), Can f 1 (dog), and Der p 1 (dust mite). ELISA signals of Der p 1 and Fel d 1 became nearly undetectable within 48-72 h (99.86% and 100% reduction, respectively), while Can f 1 showed a more gradual decrease (67.83% at 72 h). To gain structural insights, we focused on Fel d 1 as a model and performed molecular docking, 200 ns molecular dynamics simulations, and MMPBSA analysis. The subtilisin-Fel d 1 complex exhibited a large buried interface (2392 Å2) stabilized by a salt bridge, short hydrogen bonds, and hydrophobic contacts. Residue-wise energy decomposition identified key interfacial residues driving the interaction. These findings provide a structural basis for protease-mediated allergen recognition and identify B. amyloliquefaciens Ba2501 as a promising candidate for developing environmentally sustainable, enzyme-based strategies to mitigate indoor allergen exposure and improve indoor air quality. ## A set of optimized 3D-MoRSE descriptors for molecular representation - Source: npj Computational Materials (journals) - Date: 2026-09-08T00:00:00Z - Categories: Cheminformatics, Property Prediction, ADMET & Safety - Journal: npj Computational Materials - DOI: 10.1038/s41524-026-02313-5 - External ID: 7eac1f73ffc8ea4c7a7c806b0fca992bf84dab4f - Keywords: Tox21, random forest, molecular representation, molecular representations - Source URL: - Dashboard article: Abstract: Accurate early-stage prediction of molecular properties and toxicity is critical for reducing cost and attrition in drug discovery. Here, we develop and evaluate an optimized 3D-MoRSE (OPT3D) molecular representation that introduces a tunable distance-scale factor to better capture informative interatomic distance regimes. With simple traditional machine-learning models, for example, random forest (RF), OPT3D achieves an average (10 times) test root mean squared error (RMSE) of 0.942 (ESOL), 0.940 (Lipophilicity), and 2.108 (FreeSolv), and AUC of 0.824 (BACE), 0.878 (BBBP), 0.621 (SIDER), 0.828 (Tox21), and 0.717 (ToxCast). These results are competitive with recent state-of-the-art predictions that rely on more complex architectures. The prediction performance of this set of descriptors can be further improved by more advanced models: stacked ensembling further reduces regression errors and maintains strong classification performance, and the adaptive checkpointing with specialization neural model can also improve performance relative to simple traditional machine-learning models. Performed perturbation-based tests also show that the OPT3D descriptor exhibits strong robustness under moderate noise. Our results highlight that developing high-quality molecular representations is as important as model innovation, which has been intensively pursued but has yielded limited performance gains. ## A Sustainable Green Analytical Framework: Combining UV-Vis, IR, and Fluorescence Spectroscopy with Chemometric and Machine-Learning Models for the Accurate Quantification of Metformin and Sitagliptin - Source: International Journal of Recent Trends in Multidisciplinary Research (journals) - Date: 2026-09-08T00:00:00Z - Categories: Spectra & Analytical - Journal: International Journal of Recent Trends in Multidisciplinary Research - DOI: 10.59256/ijrtmr.20260605011 - External ID: 47ef6792bb7cb0cc37cf9db08bd979dd0ab4a476 - Keywords: Chemometric - Source URL: - Dashboard article: Abstract: Pharmaceutical laboratories are under growing pressure to trade solvent-heavy assay methods for greener alternatives that still meet strict quality benchmarks, and this has fed a broader move toward Green Analytical Chemistry (GAC). This review pulls together and critically evaluates published work on low-solvent, sustainable ways of assaying the antidiabetic fixed-dose combination metformin and sidesplitting covering UV-Vis, FT-IR, and molecular fluorescence spectroscopy, used both on their own and coupled with chemometric or machine learning (ML) multivariate calibration. Using a structured search across PubMed/MEDLINE, Scopus, Web of Science, ScienceDirect, and Google Scholar (2010–2026, weighted toward 2020–2026) with clearly defined inclusion and exclusion criteria, 36 primary and methodological sources were identified and compared, extending the ground covered by earlier, narrower surveys of this drug pair. Spectra collected from commercial tablets can be interpreted through Principal Component Regression (PCR), Partial Least Squares (PLS) regression, and, increasingly, artificial neural network (ANN) or support vector machine (SVM) models to disentangle overlapping signals and reduce excipient interference. Provided such methods are validated against ICH Q2(R2) and built within the ICH Q14 Analytical Quality by Design (AQbD) framework, they can match reversed-phase HPLC for linearity, precision, and accuracy while scoring noticeably better on recognized greenness metrics (AGREE, GAPI, ComplexGAPI, Eco-Scale, GEMAM). This review weighs the strengths and shortcomings of each spectroscopic approach, proposes as its central contribution a consolidated, data-fused, AQbD-compliant, green-verified framework, and considers its fit for routine batch release, stability testing, and Process Analytical Technology (PAT), including the newer wave of AI-enabled digital spectroscopy. ## A Zero-Shot Single-Point Chemical Language Model Mimics Medicinal Chemist Reasoning for Molecular Optimization and STING Inhibitor Development in Alzheimer’s Disease - Source: ChemRxiv (preprints) - Date: 2026-09-08T00:00:00Z - Categories: Design de novo - Authors: Peng Gao, Ying Qin, Zhilian Dai, Yichao Liu, Hao Wen, Jie Zhang, Songyou Zhong, Tenghuan Ge, Chen Wang, Jiawei Fu, Dan Zhang, Ning Zhi, Zhen Zhao - DOI: 10.26434/chemrxiv-2025-m82r5/v2 - External ID: 10.26434/chemrxiv-2025-m82r5/v2 - Keywords: Transformer, Molecular Optimization, binding affinity - Source URL: - Dashboard article: - Code: Abstract: Drug development relies heavily on the ability of medicinal chemists to identify productive, localized structural modifications that improve potency, selectivity, and developability while preserving favorable features of a lead scaffold. Here, we introduce the Single-Point Chemical Language Model (SpCLM), a zero-shot molecular optimization framework designed to computationally emulate this iterative medicinal chemistry paradigm. Built on a Transformer architecture, SpCLM performs controlled single-point molecular modifications encompassing common medicinal chemistry operations and generates focused, chemically interpretable analog spaces rather than large, unconstrained molecular libraries. This strategy enables efficient exploration of structure–activity relationships while maintaining close structural relationships to experimentally tractable lead compounds. Across multiple protein targets and molecular optimization tasks, SpCLM generated compact libraries typically comprising only a few hundred molecules, yet recovered 60%–80% of experimentally validated active compounds from held-out test sets that were not included during model training. Generated molecules showed substantial agreement with experimentally observed structure–activity relationships, including changes in binding affinity and functional activity, demonstrating that the model can reproduce productive chemical transformations without target-specific retraining. These results establish single-point molecular editing as an efficient strategy for translating learned medicinal chemistry knowledge into experimentally relevant molecular optimization. We further applied SpCLM to the development of stimulator of interferon genes (STING) inhibitors as potential therapeutics for Alzheimer’s disease (AD). Because aberrant activation of the cGAS–STING innate immune pathway contributes to neuroinflammatory processes associated with AD and related neurodegenerative disorders, pharmacological inhibition of STING represents an emerging therapeutic strategy. Starting from experimentally characterized STING inhibitor chemotypes, SpCLM generated focused analog series through medicinal-chemistry-like single-point modifications. Integration of model-guided generation with structure-based prioritization and experimental evaluation enabled efficient exploration and optimization of STING inhibitor chemical space, identifying analogs with improved activity and providing experimentally supported structure–activity relationships for further lead development. Together, these results demonstrate that SpCLM bridges generative molecular modeling and practical medicinal chemistry by converting broad chemical knowledge into focused, experimentally actionable structural modifications. By recovering a substantial fraction of experimentally active chemical space from only hundreds of generated candidates and enabling the optimization of therapeutically relevant STING inhibitor chemotypes, SpCLM provides a generalizable framework for reducing the experimental search space and accelerating iterative drug discovery. ## An AI-Assisted Workflow for Rapid Prioritization of FDA-Approved Drugs as HDAC3 Inhibitor Candidates for Drug Repurposing - Source: Biology (journals) - Date: 2026-09-08T00:00:00Z - Categories: Property Prediction - Journal: Biology - DOI: 10.3390/biology15181570 - External ID: 3c862c9d7699c978fdbb83d5fdb853c27054beb2 - Keywords: virtual screening, molecular descriptor, kinase - Source URL: - Dashboard article: Abstract: Epigenetic regulation through histone acetylation plays a critical role in gene expression and cancer progression. Because of its pivotal role in chromatin remodeling, Histone deacetylase 3 (HDAC3) has become a promising therapeutic target. In this study, an artificial intelligence (AI)-driven strategy was utilized to prioritize potential HDAC3 inhibitors among FDA-approved compounds to accelerate drug repurposing for cancer therapy. Existing HDAC3 inhibitors were identified in the BindingDB and were used to develop a machine learning (ML) model trained on the most potent inhibitors to identify molecular descriptor patterns associated with HDAC3 inhibition. The ML workflow then screened 1615 FDA-approved compounds, yielding 120 candidates with predicted HDAC3 inhibitory activity. Among these, known HDAC inhibitors, including romidepsin, vorinostat, and panobinostat, were selected, suggesting that the workflow can recover known HDAC inhibitors during virtual screening. Interestingly, tyrosine kinase inhibitors such as imatinib and osimertinib were also identified, indicating potential structural overlap between kinase- and HDAC3-binding pharmacophores. The analysis of the predicted docking scores also supported the prioritization results since the top 10 compounds had more negative predicted docking scores than the bottom 10 (p = 0.0074). This shows that the suggested workflow is useful for prioritizing FDA-approved compounds as potential HDAC3 inhibitors for further study. ## APPLICATIONS OF DATA ANALYSIS AND ARTIFICIAL INTELLIGENCE AGENTS IN NOVEL DRUG DISCOVERY: A CRITICAL LITERATURE REVIEW - Source: Brazilian Journal of Health Aromatherapy and Essential Oil (journals) - Date: 2026-09-08T00:00:00Z - Categories: Property Prediction, Targets & Structures, LLMs & Agents - Journal: Brazilian Journal of Health Aromatherapy and Essential Oil - DOI: 10.62435/2965-7253.bjhae.2026.69 - External ID: e41d000cfc2a0a344ff37da9f6771f37250fbb9f - Keywords: AlphaFold 3, QSAR, synthesis planning - Source URL: - Dashboard article: Abstract: The traditional drug discovery and development process is historically characterized by high attrition rates, escalating financial costs, and decade-long timelines. The emergence of artificial intelligence (AI) and machine learning (ML) has transformed this paradigm by enabling efficient navigation through vast chemical spaces and the integration of complex multi-omic datasets. This evidence-based literature review critically examines the evolution and application of computational technologies across the pharmaceutical pipeline, ranging from early expert systems like DENDRAL, computer-aided drug design (CADD), and quantitative structure-activity relationship (QSAR) modeling to AlphaFold 3 biomolecular complex predictions, computer-assisted synthesis planning (CASP), natural product bioprospecting, and autonomous multi-agent systems. Key advancements in antimicrobial screening, precision oncology, phytochemical characterization, and clinical-stage AI-generated molecules are highlighted. Finally, the translational gap is addressed, emphasizing that AI functions as an advanced decision-support framework requiring rigorous in vitro and in vivo experimental validation, wherein qualified human mediation remains indispensable for therapeutic success. ## Artificial intelligence-driven identification and mechanistic exploration of synergistic anti-aging compounds from Dengzhan Shengmai formulation - Source: PLOS One (journals) - Date: 2026-09-08T00:00:00Z - Categories: Property Prediction, ADMET & Safety - Journal: PLOS One - DOI: 10.1371/journal.pone.0356888 - External ID: 8d3e35659e32b633b97c2e14f10f8ead9f9d4381 - Keywords: molecular descriptors, ADMET - Source URL: - Dashboard article: Abstract: Aging is a complex biological process involving multiple dysregulated pathways, and synergistic compound combinations offer distinct therapeutic advantages through multi-target and multi-pathway interactions. Traditional Chinese Medicine (TCM) formulations are inherently synergistic, yet systematically identifying their active anti-aging combinations remains a major challenge. Here, we developed DeepSMCA, a deep learning-based framework integrating molecular descriptors, ADMET parameters, and protein-protein interaction (PPI) network embeddings learned via a variational graph auto-encoder (VGAE), combined with a ResNetDNN classifier and the Bliss independence model, to identify synergistic combinations of anti-aging compound from the Dengzhan Shengmai (DZSM) formulation. Trained on 914 curated compounds, DeepSMCA achieved an area under the curve (AUC) of 0.9849 on the validation set, outperforming conventional machine learning and deep learning baselines, and interpretability analysis revealed that PPI network features contributed most (59.5%) to model predictions. Chemical profiling identified 30 constituents in DZSM, from which the three top-ranked synergistic combinations (Com1–3) were validated in D-galactose (D-Gal)-induced senescent PC12 cells. All three combinations enhanced cell viability, alleviated oxidative stress, attenuated intracellular reactive oxygen species accumulation, and decreased senescence-associated β-galactosidase-positive cells by up to 54.79%. Transcriptomic analysis showed that the combinations reversed 1,001–1,037 D-Gal-induced differentially expressed genes (DEGs), which were enriched in 18 shared aging-related pathways centered on longevity regulation, FoxO, p53, and autophagy signaling. Compound-target–aging-pathway network analysis further revealed complementary target engagement among constituents. This study establishes an interpretable, proof-of-concept computational–experimental pipeline for dissecting multi-component synergy in complex formulations, providing a generalizable strategy for anti-aging drug discovery from TCM. ## Artificial intelligence-enabled cross-scale integration of traditional Chinese medicine and biomedicine for sepsis: from mechanisms to delivery - Source: Frontiers in Pharmacology (journals) - Date: 2026-09-08T00:00:00Z - Journal: Frontiers in Pharmacology - DOI: 10.3389/fphar.2026.1879407 - External ID: 08a16d42f21761992fac6e6358cafb40e0dd2977 - Keywords: graph neural networks - Source URL: - Dashboard article: Abstract: Sepsis is an organ dysfunction caused by a dysregulated host immune response to infection. Its pronounced heterogeneity and cross-scale pathological disturbances have resulted in poorly defined core therapeutic targets and inefficient drug delivery. Artificial intelligence, with its capacity for high-dimensional data integration, can serve as a data-integrative and hypothesis-generating tool, offering new avenues for exploring potential solutions to the aforementioned bottlenecks. This review summarizes recent advances in AI-driven multi-omics-based mechanistic dissection, the use of nanodelivery systems to optimize the in vivo behavior of both biomedical and botanical drugs, and AI-assisted nanocarrier design. At the mechanistic level, AI integrates multi-omics data with graph neural networks to provide computational clues for precise molecular subtyping and the identification of potential candidate targets such as S100A8/A9 and TREM-1. At the delivery level, nanocarriers help overcome the off-target toxicity of biomedical agents and the poor bioavailability of botanical drugs, while lesion acidification, high ROS levels, high MMP expression, and the EPR effect provide a biological basis for stimuli-responsive delivery. At the integration level, AI translates target information and microenvironmental parameters into carrier design parameters, offering computational support for material screening and response threshold optimization. A conceptual framework for an integrated “target recognition–drug matching–carrier design–subtype adaptation” decision model is proposed, which may inform the matching of combined biomedical and botanical drug regimens with nanodelivery systems based on patient molecular subtypes. This cross-scale integration framework may offer a reference direction for research on sepsis and other heterogeneous inflammatory diseases. ## Biocatalytic upgrading of fusel oil from bioethanol production to levulinate esters as renewable fuel oxygenates - Source: Biotechnology for Biofuels and Bioproducts (journals) - Date: 2026-09-08T00:00:00Z - Categories: Docking & Screening - Journal: Biotechnology for Biofuels and Bioproducts - DOI: 10.1186/s13068-026-02816-9 - External ID: b4d236ea9581d2814a57fe5f4b16a9183e392dc7 - Keywords: Molecular docking, chemicals, binding affinity, enzyme - Source URL: - Dashboard article: Abstract: The transition toward a sustainable bioeconomy requires efficient strategies for converting renewable biomass and industrial side streams into value-added chemicals and fuels. Fusel oil is an important coproduct of bioethanol fermentation and levulinic acid (LA), a key platform molecule derived from lignocellulosic biomass, can be upgraded into alkyl levulinates, which are promising green solvents, specialty chemicals, and fuel-related applications. In this study, a solvent-free enzymatic approach was developed for the synthesis of alkyl levulinates via lipase-catalyzed esterification of levulinic acid with bioethanol-derived fusel oil and its major alcohol components, including 2-methyl-1-butanol and 3-methyl-1-butanol, using immobilized Candida antarctica lipase B (Novozym® 435). Under optimized conditions (50 °C, LA:alcohol molar ratio 1:10, enzyme loading 10 wt% and molecular sieve addition equal to substrate mass), quantitative levulinic acid conversion and levulinate yield (~ 98–99%) were achieved within 5 h. In comparison, the secondary alcohol 2-butanol, evaluated for mechanistic comparison, reached a lower conversion (~ 70%) under otherwise identical conditions, consistent with its greater steric hindrance within the enzyme active site. Molecular docking simulations supported experimental observations by elucidating substrate–enzyme interactions and differences in binding affinity. The immobilized enzyme exhibited excellent operational stability, maintaining over 97% LA conversion over five consecutive cycles, and successful operation at 100 mL in a bioreactor highlighted the potential for further scale-up. The solvent-free system reduces process complexity, facilitates downstream processing and enhances process sustainability. Overall, this work demonstrates an efficient strategy for valorizing fusel oil side stream into high-value levulinate esters, contributing to circular bioeconomy approaches and the development of renewable fuel oxygenates and chemical platforms. ## Biopesticidal Potential of the Cannabis sativa L. Metabolites: A Denoised, Docking-Informed QSAR Model - Source: International Journal of Molecular Sciences (journals) - Date: 2026-09-08T00:00:00Z - Categories: Cheminformatics, Property Prediction, ADMET & Safety - Journal: International Journal of Molecular Sciences - DOI: 10.3390/ijms27188000 - External ID: a7364ea2bb61a6295bf1cc439d62ffc659271882 - Keywords: RDKit, QSAR, chemical diversity - Source URL: - Dashboard article: Abstract: Plant metabolites are a promising source of new biopesticides, but their chemical diversity exceeds the capacity of experimental screening. Cannabis sativa is a particularly attractive crop for discovering such compounds, although its metabolome has not been systematically evaluated for biopesticidal potential. Here, we computationally analyzed 5211 compounds annotated as C. sativa metabolites in the Cannabis Compound Database (CCD) using an integrated framework combining graph-based molecular prediction and protein–ligand interaction analysis. Initial prioritization employed a directed message passing neural network (DMPNN) trained on molecular graphs augmented with RDKit descriptors. The DMPNN predictions were integrated with a CatBoost-derived docking-consistency score based on residue-level Vina interaction terms, reducing the false-positive rate by about 60% compared with the structural model alone. Informative ligand-residue interactions were identified using a random matrix theory (RMT) framework. The DMPNN identified 1010 compounds as DMPNN-positive (score ≥0.70), indicating structural characteristics more consistent with the DS2 pesticide reference set than with the DS3 AChE-inactive reference set. Then, these compounds were filtered using annotations from the CCD to retain 44 secondary metabolites. Finally, the 44 compounds were ranked by the final ensemble score. Compared with reference pesticides, C. sativa metabolites showed higher predicted median oral LD50 values and fewer organ-specific toxicity alerts at the dataset level, although not for all endpoints. Overall, the combined structural and docking-informed workflow identified a small, chemically diverse set of high-ranking C. sativa compounds that can now be prioritized for experimental validation. ## Carbon Nanotube-Induced Magnetic Shielding Effects on 129Xe NMR from Equivariant Neural Networks - Source: ChemRxiv (preprints) - Date: 2026-09-08T00:00:00Z - Authors: Ouail Zakary, Tiia Jacklin, Perttu Lantto - DOI: 10.26434/chemrxiv.15007956/v2 - External ID: 10.26434/chemrxiv.15007956/v2 - Keywords: MLIP, graph neural networks, Equivariant, molecular dynamics - Source URL: - Dashboard article: Abstract: Equivariant graph neural networks (EGNNs) have shown success in building efficient and accurate machine learning interatomic potentials (MLIPs) for molecular dynamics (MD) and in enabling the prediction of nuclear magnetic resonance (NMR) parameters in complex systems. In this work, EGNNs are used to fine-tune an atomistic foundation model, which serves as an MLIP for large-scale molecular dynamics simulations of xenon in carbon nanotubes. In addition, EGNNs are employed to develop a machine learning model for the 129 Xe NMR magnetic shielding tensor, σ , enabling its direct prediction from MLMD snapshots. Both models are data-efficient and remain accurate well beyond their training data. Using this dual-model framework, we show that xenon encapsulated in carbon nanotubes experiences a nanotube-induced, negative 129 Xe isotropic chemical shift that varies non-monotonically with nanotube diameter and depends further on temperature and Xe loading, establishing this dual-model approach as an efficient route to interpreting confinement-driven 129 Xe NMR response. ## Chainable DataFrame Framework for Interactive Chemical Data Processing with pdChemChain - Source: ChemRxiv (preprints) - Date: 2026-09-08T00:00:00Z - Categories: Cheminformatics - Authors: Esben Jannik Bjerrum - DOI: 10.26434/chemrxiv.15008524/v1 - External ID: 10.26434/chemrxiv.15008524/v1 - Keywords: RDKit - Source URL: - Dashboard article: - Code: Abstract: Computational chemists frequently build data processing workflows as ad-hoc scripts that are difficult to reuse, share, or deploy. Yet, many of the alternative workflow platforms demand significant complexity overhead or create manual breaks in workflows involving different pieces of software. pdChemChain is an open-source Python framework that bridges interactive use in python with reuse and deployment, by treating chemical data processing chains as composable, self-serializing pandas DataFrame operations. The framework is built around three design principles: (1) every operation takes a DataFrame and returns a DataFrame, (2) operations compose via the + operator into chains that are themselves operations, and (3) all chains are self-configuring, i.e. serializable to YAML for command-line deployment without code changes. Chains are built interactively in notebooks with immediate visual feedback, then deployed as-is via CLI or as scoring components in generative design frameworks. Built on widely used RDKit and pandas, pdChemChain provides links for molecular processing, property calculation, score transformation, and multi-parameter aggregation, while remaining extensible through a minimal dataclass-based API. pdChemChain is available at https://github.com/EBjerrum/pdchemchain under the LGPL-3.0 license. ## Cloning of a novel cyanobacterial serotonin N-acetyltransferase gene CySNAT3 and its integration with transcriptomic drug repurposing in pediatric obstructive sleep-disordered breathing - Source: Frontiers in Pharmacology (journals) - Date: 2026-09-08T00:00:00Z - Categories: Docking & Screening, Targets & Structures - Journal: Frontiers in Pharmacology - DOI: 10.3389/fphar.2026.1851793 - External ID: 898708e3cb110a0e2e36ace42c40521ea34d6d6e - Keywords: molecular docking, enzyme - Source URL: - Dashboard article: Abstract: Melatonin has potential therapeutic value in pediatric obstructive sleep-disordered breathing (oSDB). However, plant-based melatonin extraction faces limitations including long growth cycles, low yield, and complex purification procedures. Cyanobacteria, as photosynthetic microorganisms, offer distinct advantages such as rapid growth, low cultivation cost, and well-established genetic manipulation tools, making them ideal chassis cells for producing high-value natural products. Meanwhile, the molecular mechanisms linking tonsillar pathology to oSDB severity remain poorly understood, and translational therapeutic targets are urgently needed. A novel cyanobacterial serotonin N-acetyltransferase gene (CySNAT3) was cloned, expressed in E. coli , and purified by Ni-NTA affinity chromatography. Its enzymatic activity was assessed by HPLC-fluorescence. Separately, differential expression analysis, WGCNA, and drug repurposing were performed on tonsillar RNA-seq data (GSE274855), followed by molecular docking to validate drug-target interactions. CySNAT3 catalyzed the conversion of serotonin to N-acetylserotonin and 5-MT to melatonin, providing an enzymatic tool for microbial melatonin production. Four high-confidence candidate drugs (sulfamethoxazole, cimetidine, diethylstilbestrol, clofibrate) were identified, with molecular docking confirming favorable binding affinities to their target proteins (CYP2C19, SLC47A1, WNT7A, SCD). Cimetidine and diethylstilbestrol have been experimentally linked to the melatonin pathway, suggesting that these four drugs may act through modulation of melatonin signaling. This study provides a new molecular foundation for pediatric oSDB treatment by integrating a novel cyanobacterial SNAT enzyme tool with transcriptome-derived drug candidates. ## Computational design, structural modeling and immune evaluation of an amastin-based multi-epitope vaccine candidate against leishmaniasis. - Source: Human immunology (journals) - Date: 2026-09-08T00:00:00Z - Categories: Docking & Screening, Free Energy & MD - Journal: Human immunology - DOI: 10.1016/j.humimm.2026.112066 - External ID: 0b8edd55afc136c05a897a88fcfbc3198eda51ed - Keywords: molecular docking, binding affinity, binding free energy, receptor, molecular dynamics - Source URL: - Dashboard article: Abstract: Leishmaniasis is a neglected tropical disease caused by protozoan parasites of the genus Leishmania. Currently, no licensed human vaccine is available and increasing drug resistance continues to compromise the disease control. The present study aimed to computationally design and evaluate a multi-epitope vaccine candidate targeting the conserved amastin protein of Leishmania donovani (UniProt ID: A5XDA6) using an integrated immunoinformatics approach. Cytotoxic T lymphocyte (CTL), helper T lymphocyte (HTL) and B cell epitopes were predicted and sequentially prioritized on the basis of HLA-binding affinity, antigenicity, allergenicity and molecular docking with their respective HLA molecules. A total of twelve epitopes (five CTL, four HTL and three B cell epitopes) were selected. Conservancy analysis demonstrated that most epitopes were highly conserved across L.donovani, L. infantum and L. major, while population coverage analysis predicted 81.61% global coverage. The selected epitopes were assembled into a 209-amino acid vaccine construct using AAY, GPGPG and KK linkers with the TLR4 agonist RS-09 as an adjuvant. The construct was predicted to be antigenic, non-allergenic, stable and soluble whereas structural validation confirmed a high-quality three-dimensional model. Molecular docking demonstrated favourable interaction with the TLR4 receptor, while discontinuous B cell epitope analysis revealed structural similarity between the vaccine construct and the native amastin protein. Normal mode analysis, molecular dynamics simulation and immune simulation further supported the structural stability and immunogenic potential of the vaccine construct. A 100 ns molecular dynamics simulation confirmed the structural stability of the vaccine-TLR4 complex and MM/PBSA and MM/GBSA binding free energy analyses further supported favourable binding affinity between the vaccine construct and TLR4. Codon optimization and in silico cloning indicated favourable expression in the Escherichia coli pET-28a (+) expression system. These computational findings suggest that the proposed multi-epitope vaccine is a promising candidate which requires further experimental validation for realizing its potential as a vaccine against leishmaniasis. ## Computational identification of isoflavonoids as novel PPAR-γ agonists for the treatment of type 2 diabetes mellitus. - Source: Technology and health care : official journal of the European Society for Engineering and Medicine (journals) - Date: 2026-09-08T00:00:00Z - Journal: Technology and health care : official journal of the European Society for Engineering and Medicine - DOI: 10.1177/09287329261481865 - External ID: db365b2a3602b07a935f3444e9d64e69c8c005ca - Keywords: molecular docking, ADMET, pharmacokinetic, docking (Glide, bioactivity, binding affinity, density functional theory, DFT, binding free energy, receptor, Molecular dynamics - Source URL: - Dashboard article: Abstract: BackgroundThe management of type 2 diabetes mellitus (T2DM) using peroxisome proliferator-activated receptor gamma (PPAR-γ) agonists is currently limited by the adverse effects of thiazolidinediones (TZDs), such as fluid retention and weight gain. This necessitates the search for safer natural alternatives.MethodsThis study applied an integrated multi-layer in silico pipeline to screen 19 isoflavonoids from the FooDB database against the PPAR-γ crystal structure (PDB: 3DZY). The workflow combined molecular docking (Glide SP), structural interaction fingerprint (SIFt) validation and density functional theory (DFT) for electronic characterization followed by ADMET profiling and prediction of PASS bioactivity. Molecular dynamics (MD) simulations (100 ns) and MM-GBSA binding free energy calculations were performed to explore dynamic stability and thermodynamic affinity of the complex.ResultsGenistin (FDB012221) and daidzin (FDB012225) were identified as lead candidates with docking score of -10.837 and -10.125 kcal/mol respectively. DFT analysis revealed that FDB012221has better kinetic stability with a large HOMO-LUMO gap of 7.80 eV, in addition, it was identified as drug-like by ADMET profiling, showing 58% human intestinal absorption and PASS analysis predicted. Molecular dynamics (MD) simulations validated the stable complex formation and FDB012221retained consistent hydrogen bonds with HIS435 (91%) and LEU325 (94%)further, the MM-GBSA values were -60 kcal/mol indicating strong thermodynamic binding.ConclusionFDB012221 is a potent PPAR-γ agonist with high binding affinity, electronic stability and a favorable pharmacokinetic profile. This isoflavonoid is a potential natural scaffold for the development of safer therapies for T2DM and is recommended for further in-vitro and in-vivo validation. ## Computational Interpretation of Functional Divergence of VOC Family Catechol Dioxygenases in Bacillus thuringiensis HHY919: Insights from Homology Modeling and Molecular Docking - Source: Catalysts (journals) - Date: 2026-09-08T00:00:00Z - Categories: Docking & Screening - Journal: Catalysts - DOI: 10.3390/catal16090811 - External ID: 08a69fea74219d564c619223e9d8232293834f82 - Keywords: Molecular Docking, virtual screening, enzyme - Source URL: - Dashboard article: Abstract: Catechol 2,3-dioxygenase (C23O) is the rate-limiting enzyme in the meta-cleavage pathway of aromatic compound degradation, yet its functional annotation within the structurally conserved VOC superfamily remains challenging due to high sequence homology. Here, we isolated a catechol-degrading strain from bovine feces and identified it as Bacillus thuringiensis HHY919 via whole-genome sequencing. Genome annotation revealed four VOC genes sharing the same COG annotation (“catechol 2,3-dioxygenase”) but divergent KO annotations (two as glyoxalases and two as C23O), suggesting functional divergence. Using homology modeling and molecular docking, we compared their binding affinities toward catechol and 11 derivatives. All four proteins showed typical meta-cleavage binding energies (−4.9 to −5.4 kcal/mol), with slightly more favorable binding than an ortho-cleavage control. Notably, gene4224 exhibited the broadest and strongest predicted affinities in virtual screening against 212 compounds, particularly for trichlorophenol (−6.0 kcal/mol) and complex natural products (qvina\_score ≤ −7.6 kcal/mol). Phylogenetic analysis and docking results jointly identified gene3355 and gene4224 as computationally prioritized C23O candidates, with gene4224 recommended as the top candidate for future enzyme engineering and bioremediation studies. This study provides a computational workflow for resolving functional ambiguity in VOC family enzymes and generating testable hypotheses for experimental validation. ## Computational small molecule drug discovery for multidrug-resistant tuberculosis: emerging targets and recent medicinal chemistry advances. - Source: Bioorganic chemistry (journals) - Date: 2026-09-08T00:00:00Z - Categories: Docking & Screening, Design de novo - Journal: Bioorganic chemistry - DOI: 10.1016/j.bioorg.2026.110483 - External ID: 1aa19d3e8714fdcca98471720cbcbf84d173475e - Keywords: molecular docking, molecular dynamics, lead optimization - Source URL: - Dashboard article: Abstract: Tuberculosis (TB) remains a major global health challenge, further intensified by the emergence of multidrug-resistant (MDR-TB) and extensively drug-resistant (XDR-TB) strains, which compromise the effectiveness of existing therapeutic regimens. Despite significant progress in TB chemotherapy, prolonged treatment duration, drug-associated toxicity, bacterial persistence, and the continuous evolution of resistance mechanisms highlight the urgent need for innovative therapeutic strategies and next-generation antitubercular agents. This review provides a comprehensive overview of the evolving landscape of TB drug discovery, covering current therapeutic approaches, mechanisms of drug resistance, and recent advances in medicinal chemistry-driven small-molecule development. The discovery and optimization of novel chemical entities, drug repurposing strategies, and target-based approaches are discussed, with particular emphasis on fragment-based drug discovery (FBDD), structure-based drug design (SBDD), and computer-aided drug design (CADD). Recent advances targeting essential and emerging mycobacterial vulnerabilities, including DprE1, InhA, ATP synthase, MmpL3, DNA gyrase, mycobacterial carbonic anhydrases (MtCAs), PanC, and other critical metabolic pathways, are highlighted to demonstrate the expanding opportunities for mechanism-guided antitubercular drug discovery. Furthermore, the integration of computational approaches, including molecular docking, molecular dynamics simulations, artificial intelligence (AI), and machine learning (ML), has significantly accelerated hit identification, lead optimization, and prediction of drug-like properties. These approaches are transforming conventional drug discovery by enabling rational design of potent molecules with improved efficacy and pharmacological profiles. However, challenges associated with target validation, drug penetration, bacterial persistence, resistance development, and clinical translation remain major obstacles. The convergence of medicinal chemistry, advanced computational technologies, and a deeper understanding of Mycobacterium tuberculosis biology represents a promising strategy for developing effective therapies against both drug-sensitive and drug-resistant TB. ## Cross-docking and redocking reveal distinct determinants of success in physics-based and AI-driven binding pose prediction in protein–ligand complexes - Source: RSC Advances (journals) - Date: 2026-09-08T00:00:00Z - Categories: Docking & Screening, Targets & Structures - Journal: RSC Advances - DOI: 10.1039/d6ra05440d - External ID: 989d2c3d94a8c5484ac2bf97ab2ef6f9939351b6 - Keywords: AutoDock Vina, AlphaFold 3, Boltz 2, DiffDock, CNN, kinase, Receptor - Source URL: - Dashboard article: Abstract: Protein–ligand pose prediction is central to structure-based drug discovery, yet the relative performance of physics-based and AI-driven methods under realistic cross-docking conditions remains insufficiently characterized. Here, we compare physics-based docking methods (AutoDock4, AutoDock Vina, and DOCK 6) with data-driven approaches, including the deep-learning model GNINA 1.3 and the diffusion-based frameworks AlphaFold 3, Boltz-2, and DiffDock. Performance was evaluated using standardised redocking and cross-docking protocols across three Alzheimer's disease targets representing distinct binding-site architectures: acetylcholinesterase (AChE; deep gorge), β-secretase 1 (BACE1; flexible flap-controlled site), and glycogen synthase kinase-3β (GSK-3β; open, solvent-exposed pocket). Physics-based methods were competitive during redocking but showed substantial performance reductions under cross-docking, whereas diffusion-based approaches generally maintained higher cross-docking accuracy. GNINA 1.3 rigid achieved an 87.7% minimum heavy-atom RMSD success rate during redocking, which decreased to 13.5% during cross-docking, whereas AlphaFold 3, Boltz-2, and DiffDock achieved cross-docking success rates of 93.1%, 89.6%, and 85.7%, respectively. AlphaFold 3 consistently outperformed Boltz-2 despite its smaller training set, suggesting that predictive performance is influenced not only by training-data volume but also by factors such as model architecture and confidence calibration. Training-overlap analysis further showed that AI-based methods retained substantial failure rates even for complexes represented in their training data, indicating that training-data overlap alone does not ensure reliable pose prediction. Under the current protocol conditions, rigid docking outperformed flexible protocols, while flexible-docking pocket volumes showed more restricted sampling relative to experimental holo structures. Among the GNINA 1.3 configurations, CNN rescoring with refinement produced the highest pose-recovery success rates, followed by CNN rescoring alone and the default Vina/empirical scoring approach in cross-docking. Receptor conformational preference was target-dependent: holo structures provided higher docking accuracy for AChE and BACE1, whose ligand-bound cavities exhibited greater structural complexity and geometric confinement that favoured pose discrimination, whereas the apo GSK-3β structure contained a larger, more solvent-exposed cavity that improved ligand accessibility and docking performance. Overall, these findings demonstrate the importance of cross-docking and training-overlap-aware evaluation for assessing docking performance under realistic conditions and provide cavity-topology-based considerations for selecting docking strategies in structure-based drug discovery. ## Cross-Scale Machine Learning for Polymer Materials: Linking Molecular Structure, Mesoscale Organization, Processing History, and Macroscopic Properties - Source: ACS Applied Polymer Materials (journals) - Date: 2026-09-08T00:00:00Z - Journal: ACS Applied Polymer Materials - DOI: 10.1021/acsapm.6c02610 - External ID: 97b374bc386f6045015cfbe0d1a3cd29e9b29fcb - Source URL: - Dashboard article: Abstract: Machine learning (ML) is increasingly used to predict polymer properties, screen candidates, and monitor manufacturing. Yet many studies remain confined to direct repeat-unit-to-property or process-to-quality correlations, while explicit state variables are often predicted only as terminal targets, treated as parallel objectives, or invoked after prediction. Here, mediator-resolved cross-scale polymer machine learning is reserved for workflows in which a condensed-state or process-state variable is measured, predicted, or physically constrained and is used to connect upstream material identity or processing history to a downstream macroscopic response or decision. We organize the literature around four links: polymer identity and statistical chain structure; condensed-state and mesoscale organization; processing history; and macroscopic response. Rather than ranking methods by headline accuracy, we compare representation fidelity, data provenance, splitting strategy, uncertainty treatment, external or experimental validation, and reproducibility. The evidence indicates that repeat-unit representations remain useful for chemical-space screening but are insufficient for sample-specific predictions unless task-relevant information on molar-mass distribution, sequence, topology, preparation, and test conditions is included. Random splits often overstate generalization, and feature-attribution methods identify model correlations rather than mechanisms. Injection molding and extrusion are examined as detailed processing cases, with other routes considered to delineate common process–structure–property requirements. We conclude with priorities for metadata standards, multimodal benchmarks, uncertainty-aware physics-integrated models, and closed-loop validation. The resulting framework distinguishes mature interpolation tasks from genuinely transferable cross-scale design workflows. ## Cryo-EM structure-based discovery of etravirine as a specific inhibitor of PRMT5/pICln protein-protein interaction for prostate cancer treatment - Source: Journal of Enzyme Inhibition and Medicinal Chemistry (journals) - Date: 2026-09-08T00:00:00Z - Journal: Journal of Enzyme Inhibition and Medicinal Chemistry - DOI: 10.1080/14756366.2026.2727844 - External ID: bbd347a40e03692455c53d852da0e00eb936d619 - Keywords: virtual screening, receptor - Source URL: - Dashboard article: Abstract: Protein arginine methyltransferase 5 (PRMT5) is overexpressed in many cancers and correlates with poor patient survival. In prostate cancer, PRMT5 cooperates with its cofactor pICln to promote tumour growth by epigenetically activating androgen receptor (AR) expression. Using a near-atomic cryo-EM structure of PRMT5/MEP50/pICln complex, we identified a previously undefined, pICln-specific protein-protein interaction (PPI) interface on PRMT5, termed P4I. Structure-based virtual screening identified the FDA-approved compound etravirine as a binder to this site. BiFC, Co-IP, and PLA assays confirmed that etravirine disrupts PRMT5/pICln interaction. A cryo-EM structure of PRMT5/MEP50/etravirine further validated on-target binding at P4I. Functionally, etravirine reduced prostate cancer cell proliferation, inhibited tumour growth, and downregulated AR and AR-V7 expression in cells and in mouse models. These results demonstrate that the unique P4I interface is a promising therapeutic target and that etravirine serves as a proof-of-concept lead compound for exploring the potential of P4I-targeted strategies in prostate cancer. ## De Novo Design of RNA Switches for Conditional Transcription Repression - Source: Angewandte Chemie International Edition (journals) - Date: 2026-09-08T00:00:00+00:00 - Authors: Cong Sun, Fan Hong - Journal: Angewandte Chemie International Edition - DOI: 10.1002/anie.1770267 - Keywords: De Novo Design - Source URL: - Dashboard article: Abstract: Controlling transcriptional synthesis of RNA provides a powerful means to design molecular circuitry and reprogram cellular behaviors. We report the computationally de novo design of a transcriptional RNA switch, termed RNA BRAKE, to precisely repress transcriptional RNA synthesis in response to a cellular RNA trigger in live cells. The design principle uses an introduced RNA trigger to regulate the co‐transcriptional folding pathway of nascently transcribed RNA BRAKEs into an RNA terminator, thereby repressing downstream RNA transcription. Without the RNA trigger, the RNA polymerase can pass through the RNA BRAKE and continue transcribing downstream genes. We validated the designed RNA BRAKE's performance by encoding the downstream gene with GFP and found that the presence of trigger RNA can significantly repress the GFP expression and GFP mRNA transcription in Escherichia coli cells. We further investigated the mechanism of RNA BRAKE design and studied its compatibility with different RNA polymerases and ribozymes for gene regulation. To demonstrate the generality of RNA BRAKE in response to cellular mRNA, we successfully developed RNA BRAKEs that enable mCherry mRNA to repress the transcription of its encoded downstream gene. This developed RNA BRAKE offers a new strategy for manipulating cellular gene expression with broad biomedical applications. ## Deciphering the Molecular Mechanisms of Apigenin in Neurodegenerative Diseases Through Network Pharmacology, Molecular Docking, and DFT Analysis - Source: Journal of Pharmaceutical Innovation (journals) - Date: 2026-09-08T00:00:00Z - Categories: Docking & Screening - Journal: Journal of Pharmaceutical Innovation - DOI: 10.1007/s12247-026-11029-4 - External ID: 852320228d397d90a5dafb3a91953422ecdab1c1 - Keywords: Molecular Docking, DFT - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Decoding Membrane Biophysics for Nucleic Acid Delivery: Bridging Bilayer Simulations and Lipid Nanoparticle Functionality - Source: ChemRxiv (preprints) - Date: 2026-09-08T00:00:00Z - Authors: Anjana Barange, Miteshkumar Moirangthem, Santosh Kumar Meena - DOI: 10.26434/chemrxiv.15008480/v1 - External ID: 10.26434/chemrxiv.15008480/v1 - Keywords: force fields, molecular dynamics - Source URL: - Dashboard article: Abstract: Lipid nanoparticles (LNPs) are a proven platform for delivering nucleic acids, evidenced by the approvals of patisiran and COVID-19 mRNA vaccines. However, the molecular mechanisms behind lipid organization, cargo encapsulation, and intracellular release remain partially understood. This review links molecular dynamics (MD) simulations of lipid bilayers to functional LNP systems through a three-level framework. (i) Structural observables: area per lipid, bilayer thickness, intrinsic curvature. (ii) Dynamic and mechanical properties: bending modulus (Kc), deuterium order parameter (SCD), lateral diffusion coefficient (DL), potential of mean force (PMF). (iii) Functional delivery outcomes: endosomal escape efficiency, RNA encapsulation, PEG shedding kinetics, cytosolic cargo release. We assess various MD frameworks, including all-atom, classical coarse-grained, and machine learning-based methods, evaluating their accuracy across the scale frontier. The pH-responsive behavior of ionizable lipids is linked to curvature stress, driving transitions from lamellar to inverted hexagonal (HII) phases, crucial for endosomal escape. We also analyze how cholesterol, helper phospholipids, and PEGylated lipids modify membrane mechanics and aid in cargo release. We highlight current limitations like fixed-protonation force fields and inadequate parameterization for new ionizable lipids. Future directions include constant-pH ## Deep learning recognises antibiotic modes of action from brightfield images - Source: Nature Communications (journals) - Date: 2026-09-08T00:00:00+00:00 - Authors: Daniel Krentzel, Kelvin Kho, Julienne Petit, Nassim Mahtal, Thomas Delerue, Max E. Huber, Agnès Zettor, Jeanne Chiaravalli, Spencer L. Shorte, Mark Brönstrup, Anne Marie Wehenkel, Ivo G. Boneca, Christophe Zimmer - Journal: Nature Communications - DOI: 10.1038/s41467-026-76355-0 - Keywords: convolutional neural - Source URL: - Dashboard article: Abstract: The antimicrobial resistance crisis urgently calls for antibiotics with novel modes of action (MoAs). While growth inhibition assays can identify antibiotic molecules, they miss promising compounds below inhibitory concentrations and cannot reveal their MoA. Imaging-based profiling of drug-treated bacteria can inform on MoA, but current approaches generally require fluorescent labelling and/or inhibitory concentrations and it remains unclear whether compounds with novel MoAs can be robustly detected. Here, we demonstrate a deep learning approach to recognise antibiotic MoAs from unlabelled images. We train a convolutional neural network to predict treatment conditions from brightfield images of Escherichia coli exposed to antibiotics covering eight MoAs. Our approach can detect drug exposure at subinhibitory concentrations and allows near-perfect MoA recognition, even when trained on only eight images per treatment condition. Previously unseen compounds are assigned to their correct MoA with good accuracy if the MoA is represented in the training data, and our method can robustly detect MoA novelty in five out of six considered MoAs, enabling microscopy-based identification of new antibiotic classes. We also achieve near-perfect MoA recognition in Klebsiella pneumoniae , suggesting applicability to other species. Our approach complements growth inhibition assays and is poised to improve the search for innovative antibiotic compounds. ## Design and computational evaluation of reversible covalent cathepsin S inhibitors using molecular docking, ADMET, molecular dynamics, MM/PBSA, and density functional theory - Source: ChemRxiv (preprints) - Date: 2026-09-08T00:00:00Z - Categories: Docking & Screening, ADMET & Safety, Targets & Structures, Free Energy & MD - Authors: Yousif Sajjad Hasan - DOI: 10.26434/chemrxiv.15008514/v1 - External ID: 10.26434/chemrxiv.15008514/v1 - Keywords: molecular docking, ADMET, density functional theory, DFT, binding free energy, molecular dynamics, MD simulations - Source URL: - Dashboard article: Abstract: Cathepsin S (CatS) is unique among lysosomal cathepsins in retaining significant proteolytic activity under extracellular conditions. Its extracellular activity has been implicated in tumour progression, chronic inflammation, autoimmune disorders, and cardiovascular diseases, making CatS an attractive target for the development of selective covalent inhibitors . In the present study, a series of 100 α-cyanoacrylamide derivatives was rationally designed and evaluated using computational approaches. The compounds were first assessed by rigid molecular docking. Top candidates were selected for pharmacokinetics study and initial 25 ns molecular dynamics (MD) simulations for further stability evaluation. Based on these results, Compounds F7 and F48 were subjected to density functional theory (DFT) calculations, covalent and non-covalent induced fit docking, followed by 100 ns MD simulations. Complex stability and binding behaviour were further examined using root mean square deviation (RMSD), root mean square fluctuation (RMSF), hydrogen-bond analysis, free-energy landscape analysis, per-residue energy decomposition, and molecular mechanics Poisson-Boltzman surface area (MM/PBSA) calculations. F7 consistently showed more favorable behaviour than F48 throughout the study. It maintained a stable binding mode, persistence interactions with key active site residues, and favorable accommodation of its alkyl substituent within the S2 binding pocket. These observations were supported by lower conformational fluctuations, stronger residue-level energetic contributions, and a more favorable MM/PBSA binding free energy (-33.47 kcal/mol) and than F48 (-27.72 kcal/ mol). The findings provide a molecular basis for the design of extracellularly targeted covalent CatS inhibitors and identify F7 as a suitable scaffold for further experimental investigations. ## DFT and molecular docking insights into barium- and polyacrylic acid-integrated nickel selenide for efficient dye reduction and biological applications - Source: Nanoscale Advances (journals) - Date: 2026-09-08T00:00:00Z - Categories: Docking & Screening - Journal: Nanoscale Advances - DOI: 10.1039/d6na00380j - External ID: d1887bc1d73df13d7c3392e9c91fef07deae0f26 - Keywords: molecular docking, DFT - Source URL: - Dashboard article: Abstract: The escalating discharge of industrial waste has hampered access to clean and potable water. This study aims to explore the potential use of nanomaterials for eliminating organic pollutants and harmful microbes such as Escherichia coli (E. coli) from wastewater. Nickel selenide (NiSe) doped with varying concentrations (2% and 4%) of barium (Ba2+) and a fixed concentration (3%) of polyacrylic acid (PAA) was synthesized via a coprecipitation route. An array of cutting-edge characterization techniques were employed to analyze the crystallographic structure, morphological features, elemental composition, molecular vibrations and optical properties of the synthesized NSs. 4% Ba/PAA-doped NiSe displayed significant catalytic activity (CA) against rhodamine B (RhB) dye, exhibiting a removal efficiency of 85.37% under neutral conditions, attributed to energy level adjustments and morphological transformation caused by dopants (PAA and Ba2+). Furthermore, a potent bactericidal effect was shown against E. coli, with an inhibition zone of 6.90 mm. The inhibitory effect of Ba/PAA-doped NiSe NSs on DNA gyrase in E. coli was elucidated by molecular docking investigations, which validated their bactericidal activity. The DFT investigation, including MESP, HOMO, and LUMO analyses, provided significant insights into the electronic structure and active compound reactivities. ## Distinct Lipid Binding Dynamics and Sodium Ion Coordination in the TRAP Transporter Hi SiaQM Revealed by Multiscale Molecular Dynamics Simulations - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-08T00:00:00+00:00 - Categories: Cheminformatics - Authors: Senwei Quan, James S. Davies, Rachel A. North, Kelsi R. Hall, Liam S. Turk, Michael C. Newton-Vesty, Renwick C. J. Dobson, Jane R. Allison - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c01356 - Source URL: - Dashboard article: Abstract: Tripartite ATP-independent periplasmic (TRAP) transporters are substrate-binding protein dependent secondary transporters that leverage ion gradients to facilitate substrate translocation across bacterial and archaeal membranes. Although the structure of HiSiaQM─a membrane-embedded TRAP subunit from the human pathogen Haemophilus influenzae ─has recently been determined, its transport mechanism is not fully understood. Given the likely importance of lipid interactions and ion coordination dynamics to the transport cycle, we applied multiscale molecular dynamics simulations to investigate these two complementary aspects of HiSiaQM function. Our simulations validated an experimentally identified lipid-binding site (BS1) and provided insight into its lipid-binding dynamics. BS1 is preferentially occupied by the anionic POPG, which is stabilized through persistent hydrogen bonds involving residues H100, R104, and R181. Additionally, our simulations revealed a novel, less stable lipid-binding site (BS2) that is selectively occupied by POPE. We suggest distinct lipid-binding behaviors for POPG and POPE, with POPG forming more persistent interactions at BS1 and POPE engaging more transiently at BS2; however, these remain to be tested experimentally. Analysis of sodium ion coordination at two known binding sites (Na1 and Na2) revealed contrasting dynamics: Na1 exhibited stable ion retention through backbone carbonyl coordination, while Na2 displayed lower stability due to involvement of the flexible side chain of T561. Geometric tunnel analysis identified possible sodium-exit pathways, supporting a working hypothesis that Na1 may act as a stable anchoring site and Na2 may function as a transient ion-exchange point during the transport cycle. Collectively, our findings provide mechanistic insights into how specific lipid interactions and sodium ion coordination could modulate HiSiaQM transporter function. ## DL\_FFLUX Refactored: Accelerating Quantum Chemical Topology Simulations With Optimized Parallelization - Source: Journal of Computational Chemistry (journals) - Date: 2026-09-08T00:00:00+00:00 - Authors: Mohamadhosein Nosratjoo, Michael K. Bane, Paul L. A. Popelier - Journal: Journal of Computational Chemistry - DOI: 10.1002/jcc.70501 - Keywords: machine learning potential, Molecular dynamics - Source URL: - Dashboard article: Abstract: Molecular dynamics simulations are only as useful as they are both accurate and efficient. Here, we introduce an upgraded version of the simulation code DL\_FFLUX that dramatically extends accessible time scales by combining a more efficient core algorithm with OpenMP and MPI parallelization. DL\_FFLUX is a unique, state‐of‐the‐art polarisable machine learning potential (MLP) that combines Quantum Chemical Topology (QCT) and machine learning (ML) and relies on DL\_POLY as the driver for molecular dynamics simulations. This new implementation of DL\_FFLUX delivers a massive performance boost, making simulations 6.8 to 49 times faster without compromising accuracy. This performance boost of DL\_FFLUX is reported for condensed‐phase simulations of water and ethanol and for single‐molecule simulations of 6 small to medium‐sized molecules. Moreover, we compared the optimized version of DL\_FFLUX with other MLPs on the MD17 dataset for four molecules. DL\_FFLUX is the fastest: depending on the system, it outperforms the second‐fastest MLP by a factor of 1.8–9.6. ## Dynamic vacancy redistribution shifts Pd nanoparticle coalescence from orientation-gated to defect-enabled on reduced CeO2(110) - Source: ChemRxiv (preprints) - Date: 2026-09-08T00:00:00Z - Categories: ML Potentials - Authors: Shulin Wang, Jilun Song, Liang Cao - DOI: 10.26434/chemrxiv.15005781/v2 - External ID: 10.26434/chemrxiv.15005781/v2 - Keywords: MLIP, density functional theory, DFT, ab initio, molecular dynamics - Source URL: - Dashboard article: Abstract: Sintering deactivates supported catalysts, yet how mobile support defects couple to crystallographic transport anisotropy during nanoparticle coalescence remains poorly understood. Here, we develop a ternary Ce–O–Pd machine-learning interatomic potential (Ce–O–Pd MLIP) by fine-tuning a pre-trained large atomic model and validate it against density functional theory (DFT) calculations and ab initio molecular dynamics (AIMD). Large-scale deep potential molecular dynamics (DPMD) simulations resolve early-stage coarsening of Pd 62 /Pd 287 nanoparticle pairs on CeO 2–x (110) and reveal direction-selective coalescence: \[11̅0\]-aligned pairs exhibit higher probabilities and shorter onset times than \[001\]-aligned pairs. CI-NEB calculations identify complementary transport anisotropies, with Pd adatom diffusion favored along \[11̅0\] and oxygen-vacancy migration favored along \[001\], enabling dynamic defect redistribution within the interparticle region. Statistical trajectory analysis shows that increasing the interparticle oxygen-vacancy fraction from 10% to 30% raises coalescence probability and accelerates bridge and neck nucleation, including along the kinetically less favorable \[001\] alignment. Time-resolved metrics further show that transient low-coordinated bridge-Pd intermediates precede complete particle merging. Together, these results establish defect–transport coupling as a microscopic mechanism for direction-selective sintering on reduced ceria and suggest that suppressing vacancy enrichment and connectivity in interparticle gaps can improve the stability of oxide-supported metal nanoparticles. ## Electronic-Structure Determinants of Flavonoid Inhibition at the Molybdenum Cofactor of Xanthine Oxidase: A QM/MM Study - Source: ACS Omega (journals) - Date: 2026-09-08T00:00:00+00:00 - Categories: Docking & Screening - Authors: Mei-Ling Li, Cheng-Hong Hsieh, Shang-Ming Huang, Kuo-Chiang Hsu - Journal: ACS Omega - DOI: 10.1021/acsomega.6c08309 - Keywords: molecular docking, B3LYP - Source URL: - Dashboard article: Abstract: Flavonoid-mediated inhibition of xanthine oxidase (XO) is well established at the empirical level; however, while molecular docking, MD, and MM-PBSA/MM-GBSA methods have been widely applied, the quantum-mechanical origin of the structure–activity relationship at the molybdenum cofactor (MoCo) remains unresolved. Hybrid QM/MM calculations (B3LYP/def2-SVP) on five dietary flavonoids bound to the MoCo active site of bovine XO (50 snapshots total) revealed that all five exhibit net charge redistribution at the MoCo interface, with the magnitude varying nearly 9-fold. The charge-redistribution descriptor |CTlig| correlated with Ki (r = – 0.922, p = 0.026; Spearman ρ = – 1.000), and Hirshfeld charge analysis independently confirmed the Mulliken-derived ranking (r = 0.942, p = 0.017), while representative def2-TZVP calculations preserved the principal compound hierarchy (Spearman ρ = 0.900), supporting robustness to both charge partitioning and basis-set effects. Frontier orbital analysis links the observed structure–activity relationship to the saturated C2–C3 bond of naringenin, which interrupts π-conjugation and limits electronic communication toward the MoCo environment. The electrostatic interaction energy further correlated with Ki (r = – 0.890, p = 0.043), whereas total interfragment contact density showed no correlation, indicating that within this five-compound series, inhibitory potency was more closely associated with charge-redistribution magnitude and electrostatic interaction than with the interfragment contact-density descriptor. ## Enantiomer-Resolved Biological Profiling and Chiral Developability Assessment of Novel Naphthylethyl Thioureas - Source: Pharmaceutics (journals) - Date: 2026-09-08T00:00:00Z - Categories: Docking & Screening - Journal: Pharmaceutics - DOI: 10.3390/pharmaceutics18091129 - External ID: e702f28db245892a9bde0c9ee2146510b415e306 - Keywords: molecular docking - Source URL: - Dashboard article: Abstract: Background/Objectives: Stereochemistry can substantially influence the biological activity and pharmaceutical properties of small molecules. This study investigated the stereochemistry-dependent biological activity and chiral recognition of novel naphthylethyl thiourea enantiomers using an integrated biological, chromatographic, and computational approach. Methods: Ten pairs of naphthylethyl thiourea enantiomers were synthesized from enantiomerically pure precursors and characterized. Antiproliferative and antibacterial activity, chiral HPLC behavior on polysaccharide- and protein-based stationary phases, molecular docking, and multivariate relationships were evaluated. Results: Target-dependent enantioselectivity and substituent-dependent activity patterns were observed. All enantiomeric pairs were chromatographically distinguished under at least one condition, and selected compounds showed selector- and mobile-phase-dependent changes in elution order. AGP-based chromatography showed stereoselective recognition for a subset of compounds, while computational analyses provided complementary support for the experimental trends. Conclusions: Stereochemistry and substitution jointly influence biological and chromatographic behavior in this thiourea series. Integrated biological, chromatographic, and computational profiling provides a useful framework for early enantiomer-resolved developability assessment. ## Epitope-Guided Two-Step Enzymatic Hydrolysis for Targeted Control of the Immunoreactivity of the Major Sichuan Pepper Allergen Zan b 2 - Source: Journal of Agricultural and Food Chemistry (journals) - Date: 2026-09-08T00:00:00Z - Categories: Docking & Screening - Journal: Journal of Agricultural and Food Chemistry - DOI: 10.1021/acs.jafc.6c08943 - External ID: 8fe10bbaae89a21c120ad2cc5b56ba826071d6f7 - Keywords: molecular docking, molecular dynamics - Source URL: - Dashboard article: Abstract: Sichuan pepper (Zanthoxylum bungeanum) is widely consumed in China, but its allergenicity raises food-safety concerns. In this study, the 11S globulin allergen Zan b 2 was identified as the major allergen and characterized through structural and immunological analyses. Integrated immunoinformatics, molecular docking, and molecular dynamics simulations were used to predict B-cell epitopes and screen proteases capable of specifically cleaving these regions. Prediction results indicate that proteinase K targets the following epitopes: Glu13, Thr100, Tyr149, Glu164, Glu229, Glu231, Leu239, Glu241, Thr245, Val328, Glu368, Ala381, and thermolysin targets Arg182, Lys247, Ser273, Asn294, Gln332, and Arg380. Experimental validation using enzymatic hydrolysis, SDS-PAGE, indirect ELISA, LC–MS/MS, and proteomics showed that combined treatment with proteinase K and thermolysin markedly reduced the IgE-binding activity. These findings demonstrate that simulation-assisted targeted enzymatic hydrolysis is an effective strategy for reducing Zan b 2 immunoreactivity and supports the development of hypoallergenic Sichuan pepper products. ## Explainable ML force-field for evaluating protein–ligand binding energy using SO3LR - Source: ChemRxiv (preprints) - Date: 2026-09-08T00:00:00Z - Categories: Docking & Screening, Free Energy & MD - Authors: Hamza Agha, Sergio Suárez-Dou, Adil Kabylda, Alexandre Tkatchenko, Andrea Volkamer - DOI: 10.26434/chemrxiv.15008482/v1 - External ID: 10.26434/chemrxiv.15008482/v1 - Keywords: Machine learned force fields, equivariant, force fields, molecular dynamics - Source URL: - Dashboard article: Abstract: Accurate estimation of protein–ligand binding affinities remains central to structure-based drug design. Classical force fields neglect important quantum-mechanical effects, while quantum mechanical and semi-empirical methods remain impractical for large biomolecular systems. Machine-learned force fields (MLFFs) offer an alternative, but their performance may be inflated by training on protein–ligand complexes. Here, we introduce SO3LR-SF, a physics-based scoring function built on the pretrained SO3LR MLFF, combining an equivariant message-passing network for semi-local interactions with physically motivated terms for short-range repulsion, electrostatics, and long-range dispersion. Built on an MLFF that was not trained on protein–ligand complexes, SO3LR-SF provides a stringent test of transferability. On PLA15 (15 active-site models), SO3LR-SF attains a relative interaction error of 5.8% against DLPNO-CCSD(T) references, the lowest of all tested methods. On the FEP benchmark (8 targets, 264 ligands), SO3LR-SF achieves an average Spearman correlation of 0.51, on par with the best semi-empirical quantum-mechanical method tested (0.47, GFN-FF) and with MMGBSA (0.44), and surpassing Glide (0.27). On the Wang dataset (8 targets, 199 ligands), it reaches 0.51, approaching the 0.60 average of molecular dynamics-based free-energy methods at a fraction of their cost. Each complex is scored in 1–10 seconds on 12 CPU cores, an 80- to 300-fold speedup over comparable MLFFs. An 8 Å trimming protocol reduces runtime threefold without loss of accuracy, and a restrained geometry optimization recovers ranking accuracy for targets with strained input structures. A multi-level explainability framework adds energy decomposition, per-atom energy contributions reported either for ligand or as 2D protein–ligand interaction maps, and 3D binding site visualization. We identified two pocket descriptors that flag applicability before scoring, e.g. polar solvent accessible surface area (SASA) ratio correlating strongly (r = 0.81) with performance. Together, SO3LR-SF presents a fast, interpretable, and competitive scoring function for drug discovery that requires no domain-specific training data. ## Feasibility of longitudinal in vivo monitoring of pulmonary disease progression in mouse models using laboratory-based x-ray dark-field CT - Source: European Radiology Experimental (journals) - Date: 2026-09-08T00:00:00Z - Journal: European Radiology Experimental - DOI: 10.1186/s41747-026-00789-w - External ID: a11bb131a98a1dbd1d82d6dfe876ac77d12e1851 - Source URL: - Dashboard article: Abstract: Early alterations in pulmonary microstructure are central to the onset and progression of chronic obstructive pulmonary disease (COPD) and acute lung injury, yet these changes remain undetectable with conventional attenuation-based computed tomography (CT). Dark-field computed tomography is sensitive to small-angle x-ray scattering generated by intact alveolar structures; however, all prior in vivo studies have been cross-sectional and pseudo-longitudinal, precluding direct observation of disease evolution within the same subject. A longitudinal, microstructure-resolved imaging approach is needed to capture true disease trajectories, reduce inter-subject variability, and support preclinical therapeutic development. We developed a dedicated dark-field CT imaging system and integrated workflow, including dose-optimized acquisition, a mouse-specific fixation device, motion compensation, pseudo-dark-field suppression, and deep learning–based three-dimensional lung segmentation, to support repeated imaging over 12 weeks in healthy, inflammatory-injury, and COPD mouse models. The dark-field coefficient (µd) showed distinct, model-specific temporal trajectories and appeared to reflect microstructural changes when attenuation (μ) remained stable. It exhibited pronounced left–right heterogeneity in the inflammatory-injury model and markedly different trajectories between two COPD mice under an identical protocol, underscoring inter-subject variability that the within-subject design captured. Overall, µd changed earlier than μ, and findings were consistent with terminal histology. Our results suggest the feasibility of long-term in vivo dark-field CT in preclinical lung studies. The dark-field coefficient shows promise as a potential noninvasive biomarker for early pulmonary damage, quantitative disease assessment, and therapy monitoring, supporting further investigation of longitudinal dark-field imaging in lung pathology and drug development. Question First laboratory-based longitudinal in vivo dark-field CT in mouse lung disease models over 12 weeks. Findings Dark-field signal showed more pronounced temporal variation than attenuation signal, with distinct model-specific trajectories. Relevance statement These findings establish a methodological foundation for longitudinal preclinical imaging of lung microstructure, which may support future translational research and the development of time-resolved imaging strategies for pulmonary disease. Question First laboratory-based longitudinal in vivo dark-field CT in mouse lung disease models over 12 weeks. Findings Dark-field signal showed more pronounced temporal variation than attenuation signal, with distinct model-specific trajectories. Relevance statement These findings establish a methodological foundation for longitudinal preclinical imaging of lung microstructure, which may support future translational research and the development of time-resolved imaging strategies for pulmonary disease. ## FedMediFormer-XAI: Federated Multimodal Transformers with Diffusion Augmentation and Graph-Based Drug Recommendation for Diabetes. - Source: Journal of visualized experiments : JoVE (journals) - Date: 2026-09-08T00:00:00Z - Journal: Journal of visualized experiments : JoVE - DOI: 10.3791/73113 - External ID: 475cc297b12537d1dc44f954afd2f4a55ae59521 - Keywords: Transformers, transformer, Graph Neural Networks, GNNs, GNN, diffusion models - Source URL: - Dashboard article: Abstract: Diabetes management faces a number of obstacles, such as fragmented healthcare data, privacy concerns, poor explainability, and the lack of personalized therapeutic guidance. This research work introduces FedMediFormer-XAI, a unified and proper framework that incorporates federated learning, multimodal transformers, diffusion-based data augmentation, Graph Neural Networks (GNNs) for drug recommendation, and Explainable Artificial Intelligence (XAI) for diabetes intelligence. The system utilizes diverse healthcare data types, e.g., clinical records, population health indicators, continuous glucose monitoring data, retinal fundus images, wearable sensor measurements, and pharmacological information. To generate synthetic samples and address class imbalance, diffusion models are used, and multimodal transformer architectures are employed to learn relationships among very different data sources. Federated learning enables collaborative training of models in a privacy-preserving manner without sharing raw patient data. The GNN component captures patient-drug and drug-drug interactions for personalized drug recommendations. Explainability methods such as SHapley Additive exPlanations (SHAP), attention visualization, Integrated Gradients, and counterfactual reasoning give transparent interpretations of prediction and recommendation results. In the representative implementation, the proposed framework achieved an accuracy of 94.2%, precision of 93.1%, recall of 92.8%, F1-score of 92.9%, Matthews Correlation Coefficient (MCC) of 0.88, and an Area Under the Receiver Operating Characteristic Curve (AUC-ROC) of 0.96. The graph-based recommendation module achieved precision = 0.89, recall = 0.84, and Normalized Discounted Cumulative Gain (NDCG) = 0.91. The protocol provides a structured approach for integrating heterogeneous healthcare data using multimodal transformers, federated learning, diffusion-based augmentation, graph-based recommendation, and explainable artificial intelligence. The reported computational results demonstrate the framework's potential for diabetes prediction and personalized medication recommendation, while the federated design supports decentralized data handling. A prospective multicenter clinical evaluation is required to establish clinical utility, generalizability, and real-world applicability. ## Flow-Guided Chemical Language Modeling for Linker Design in Reticular Chemistry - Source: Journal of the American Chemical Society (journals) - Date: 2026-09-08T00:00:00+00:00 - Authors: Dhruv Menon, Vivek Singh, Xu Chen, Mohammad Reza Alizadeh Kiapi, Ivan Zyuzin, Hamish W. MacLeod, Nakul Rampal, William Shepard, Omar M. Yaghi, David Fairen-Jimenez - Journal: Journal of the American Chemical Society - DOI: 10.1021/jacs.6c06920 - Keywords: inverse design - Source URL: - Dashboard article: Abstract: Reticular chemistry has enabled the synthesis of tens of thousands of metal–organic frameworks (MOFs), yet the discovery of new materials still relies largely on intuition-driven linker design and iterative experimentation. As a result, researchers explore only a small fraction of the vast chemical space accessible to reticular materials, limiting the systematic discovery of frameworks with targeted properties. Here, we introduce NexerraR1, a building-block chemical language model that enables inverse design in reticular chemistry through targeted generation of organic linkers. Rather than generating complete frameworks directly, Nexerra operates at the level of molecular building blocks, preserving the modular logic that underpins reticular synthesis. The model supports both unconstrained generation of low-connectivity linkers and scaffold-constrained design of symmetric multidentate motifs compatible with predefined nodes and topologies. We further combine linker generation with flow-guided distributional targeting to steer the generative process toward application-relevant objectives while maintaining chemical validity and assembly feasibility. The generated linkers are subsequently assembled into three-dimensional frameworks and structurally optimized to produce candidate materials compatible with experimental synthesis. Using NexerraR1, we validate this strategy by rediscovering known MOFs and by proposing the experimental synthesis of a previously unreported framework, CU-525, generated in silico. Together, these results establish a controllable building-block-level design framework for reticular chemistry in which chemical language modeling enables the direct translation from computational design to synthesizable frameworks. ## Gamma Irradiation Effects on Fatty Acid Profiles, Adhesion, Plasmid Stability, and Secreted Proteins in Shigella: In Silico Insights into Post-Irradiation Virulence - Source: Microorganisms (journals) - Date: 2026-09-08T00:00:00Z - Journal: Microorganisms - DOI: 10.3390/microorganisms14091986 - External ID: ea5fbb04371c3b320701947099e661152db0a2aa - Keywords: molecular docking - Source URL: - Dashboard article: Abstract: This study investigated the adaptive mechanisms of Shigella sonnei ATCC25931 in response to gamma irradiation stress (0.5 and 1 kGy), with a specific focus on the involvement of fatty acids (FAs) in membrane lipid composition remodeling, adherence, and extracellular proteins. The results demonstrated a notable enhancement in cell hydrophobicity and adherence to KB cells following irradiation, although the invasion rate exhibited a decline from 14% to 2.35% at 1 kGy. Furthermore, gamma irradiation induced notable alterations in fatty acid (FA) composition, characterized by a pronounced reduction in the unsaturated/saturated FA ratio. Additionally, the plasmid profiles revealed the loss of several original plasmids after gamma irradiation. The alterations in extracellular proteins were observed through SDS-PAGE, which demonstrated a reduction in expression at 1 kGy. Furthermore, a molecular docking analysis indicated that fatty acids play a role in inhibiting VirF, a key regulator of Shigella virulence. The least effective inhibitor was capric acid, while linoleic and gamma-linolenic acids formed the most stable inhibitory complexes, which prevented VirF from activating the virulence system. These alterations in fatty acids, plasmids, and extracellular proteins constitute adaptive responses to irradiation-induced stress. ## Generative Active Learning for Molecular Design with REINVENT: Balancing Binding Affinity and Synthetic Accessibility - Source: Journal of Chemical Theory and Computation (journals) - Date: 2026-09-08T00:00:00+00:00 - Authors: Marco Klähn, Hannes H. Loeffler, Shunzhou Wan, Alexey Voronov, Xibei Zhang, Agastya P. Bhati, Peter V. Coveney - Journal: Journal of Chemical Theory and Computation - DOI: 10.1021/acs.jctc.6c01017 - Keywords: REINVENT, multi objective optimization, Binding Affinity - Source URL: - Dashboard article: Abstract: We present a generative active learning (GAL) framework for molecular design that integrates the generative AI platform REINVENT with physics-based free-energy estimation via ESMACS, specifically addressing synthetic tractability. In a previous study we have shown that iterative generation and optimization of molecules for protein binding affinity can be effectively achieved. For drug discovery, however, molecules need to be synthesized for experimental validation. Molecules with unclear synthetic routes will be de-prioritized regardless of their predicted binding affinity. Here, we address this by incorporating additional synthesizability constraints into the computational design process, framing the problem as a multi-objective optimization task. We show that it is possible to simultaneously optimize candidate molecules for both binding affinity and synthetic accessibility. Our results show that integrating synthesizability prediction into physics-based GAL workflows enables the efficient design of compounds that are at once chemically diverse as well as predicted to be strong binders and synthetically tractable, demonstrating efficient and practical computational drug design. ## Generative Active Learning for Molecular Design with REINVENT: Balancing Binding Affinity and Synthetic Accessibility - Source: Journal of Chemical Theory and Computation (journals) - Date: 2026-09-08T00:00:00Z - Journal: Journal of Chemical Theory and Computation - DOI: 10.1021/acs.jctc.6c01017 - External ID: a136d64cb286dfe2beabd152f657795c462b6ede - Keywords: REINVENT, multi objective optimization, Binding Affinity - Source URL: - Dashboard article: Abstract: We present a generative active learning (GAL) framework for molecular design that integrates the generative AI platform REINVENT with physics-based free-energy estimation via ESMACS, specifically addressing synthetic tractability. In a previous study we have shown that iterative generation and optimization of molecules for protein binding affinity can be effectively achieved. For drug discovery, however, molecules need to be synthesized for experimental validation. Molecules with unclear synthetic routes will be de-prioritized regardless of their predicted binding affinity. Here, we address this by incorporating additional synthesizability constraints into the computational design process, framing the problem as a multi-objective optimization task. We show that it is possible to simultaneously optimize candidate molecules for both binding affinity and synthetic accessibility. Our results show that integrating synthesizability prediction into physics-based GAL workflows enables the efficient design of compounds that are at once chemically diverse as well as predicted to be strong binders and synthetically tractable, demonstrating efficient and practical computational drug design. ## Homology modeling and molecular docking of selected agricultural pesticides to Halyomorpha halys (Stål) (Hemiptera: Pentatomidae) acetylcholinesterase - Source: Mustafa Kemal Üniversitesi Tarım Bilimleri Dergisi (journals) - Date: 2026-09-08T00:00:00Z - Journal: Mustafa Kemal Üniversitesi Tarım Bilimleri Dergisi - DOI: 10.37908/mkutbd.1817340 - External ID: 69992a19ebd9deca1453171438fb446ff30a6dcc - Keywords: molecular docking - Source URL: - Dashboard article: Abstract: Halyomorpha halys (H. halys) is an invasive agricultural pest that causes significant economic losses and negatively impacts human health due to its and global distribution. The limitations of current control methods and increasing pesticide resistance necessitate the development of new targeted strategies. In this study, acetylcholinesterase (AChE), which is critical for the insect nervous system, was targeted. Targeting AChE is considered an approach to developing pesticides that exhibit high efficacy and selectivity. Since the crystal structure of H. halys AChE has not been resolved, a homology model was constructed using the amino acid sequence retrieved from the NCBI database and the Drosophila melanogaster AChE structures (PDB IDs: 6XYS and 6XYY) as templates. Molecular docking analyses were performed on the created model for the selected insecticides. Compounds with binding energies below −8.0 kcal/mol and exhibiting similar interaction motifs were prioritized as candidates. Specifically, hydrogen bonds with Tyr/Thr residues and π–π and π-alkyl interactions in Trp/Tyr aromatic pockets were identified as common binding motifs. These findings suggest that the identified compounds may serve as structural starting points for the development of novel pesticides targeting AChE. However, these results should be validated by in vitro and in vivo studies. ## Identification of the Acetamide-Chalcone Derivative That Selectively Inhibits ABCG2 Transport Activity - Source: ACS Omega (journals) - Date: 2026-09-08T00:00:00+00:00 - Categories: Docking & Screening - Authors: Arthur Henrique Gomes de Oliveira, Eli Silveira Alves Ducas, Thales Kronenberger, Jean Carlos Pereira Sousa, Bruna Estelita Ruginsk, Katalin Goda, Fabiane Gomes de Moraes Rego, Geraldo Picheth, Vivian Rotuno Moure, Pablo José Gonçalves, Glaucio Valdameri - Journal: ACS Omega - DOI: 10.1021/acsomega.6c07380 - Keywords: Molecular docking, IC50, molecular dynamics - Source URL: - Dashboard article: Abstract: Multidrug resistance (MDR) remains a major obstacle to effective cancer chemotherapy, often driven by the overexpression of ABC transporters such as ABCG2. In this study, we identified the synthetic chalcone derivative, B5, as a selective ABCG2 inhibitor. Among the 20 chalcone analogs evaluated, B5, bearing an acetamide group at R1 and an ethoxy substituent at R2, was the only compound that produced more than 50% inhibition of ABCG2-mediated Hoechst 33342 transport at 10 μM, while showing no activity against ABCB1. B5 exhibited an IC50 of 2.7 μM for ABCG2 inhibition, consistent with the low-micromolar potency reported for chalcone-based ABCG2 modulators. Furthermore, B5 effectively reversed ABCG2-mediated resistance to the chemotherapeutic agent SN-38 in ABCG2-overexpressing cells, thereby restoring drug sensitivity. Notably, B5 was noncytotoxic at concentrations up to 5 μM and did not exhibit differential cytotoxicity between parental and ABCG2-overexpressing cells, suggesting that it was not efficiently transported under the experimental conditions. ATPase assays revealed that B5 stimulated, rather than inhibited, ABCG2 ATPase activity, a behavior previously reported for chemically distinct ABCG2 inhibitors. Molecular docking and molecular dynamics simulations suggested that B5 bound within the central inhibitor-binding cavity of ABCG2, forming π–π interactions with Phe439 and hydrophobic contacts with Val546, although with a lower interaction frequency than the reference inhibitor Ko143. This binding mode, together with the physicochemical properties of B5, suggested that the electron-donating acetamide group enhanced its inhibitory activity, potentially by facilitating hydrogen-bonding interactions within the binding pocket. Collectively, these findings identified B5 as a promising lead compound for the development of selective ABCG2 inhibitors and provided structural insights to support the rational design of more potent analogs capable of overcoming MDR in cancer. ## Machine Learning‐Guided Elucidation of Spacer‐Dependent Crystallization Pathways in Quasi‐2D Perovskites: Linear vs. Branched Spacer Cations - Source: Angewandte Chemie International Edition (journals) - Date: 2026-09-08T00:00:00+00:00 - Authors: Minwook Jeon, Jimin Seo, Dongyup Shin, Jin Ho Bang, Junsang Cho - Journal: Angewandte Chemie International Edition - DOI: 10.1002/anie.6439085 - Keywords: molecular descriptors, molecular features - Source URL: - Dashboard article: Abstract: Quasi‐two‐dimensional (2D) metal halide perovskites have emerged as a structurally robust, electronically tunable platform for optoelectronic applications. The precise modulation of the n ‐phase distribution in quasi‐2D metal halide perovskites remains a critical challenge for tailoring the resulting optoelectronic properties. However, molecular descriptors governing spacer‐dependent crystallization and n ‐phase distribution remain elusive. Herein, we present an integrated framework utilizing machine learning (ML) to identify the molecular descriptors of spacer cations that control structural evolution across the 2D–3D perovskite landscape. By combining multiple ML regression algorithms with SHapley Additive exPlanations (SHAP) analysis, we identified the melting point and rotatable bond count of spacer cations as the most influential molecular descriptors. These molecular features are associated with aggregational enthalpy and conformational flexibility, which are indicative of the thermodynamic assembly and kinetic diffusion processes that determine the average layer thickness. The ML‐integrated frameworks discussed in this study establish a predictive platform that bridges the gap between molecular descriptors and target optoelectronic properties, further providing a data‐driven route toward the rational design of lower‐dimensional perovskites. ## MAGNN-DTA: Cross-Modal Mutual Attention Graph Neural Networks for Drug-Target Affinity Prediction. - Source: IEEE transactions on computational biology and bioinformatics (journals) - Date: 2026-09-08T00:00:00Z - Categories: Cheminformatics - Journal: IEEE transactions on computational biology and bioinformatics - DOI: 10.1109/TCBBIO.2026.3732146 - External ID: 2447ae773ecb57486a314c72d04f1a5cb24b6d84 - Keywords: Graph Neural Networks, graph neural network - Source URL: - Dashboard article: - Code: Abstract: Drug-Target Affinity (DTA) prediction is a crucial step in discovering and developing new pharmaceuticals. However, existing approaches remain limited in modeling multimodal features and capturing cross-modal interactions, often failing to fully characterize complementary information of drugs and proteins at the sequence, structural, and topological levels. To address these limitations, we propose MAGNN-DTA, a cross-modal mutual attention graph neural network framework for DTA prediction. First, MAGNN-DTA adopts a dual-path feature extraction module to extract sequence features and graph features of drug and protein. For drug representation, a Conformer block is used to capture long-range dependencies and local functional patterns, while a Graph Isomorphism Network (GIN) and multi-head attention are used to extract local and global features from molecular graphs. The drug features are further refined via a Kolmogorov-Arnold Network (KAN) to improve representational capacity and interpretability. For protein representation, sequence and structural information are integrated through a multi-scale deep convolutional network and a gated graph attention network (Gated-GAT) derived from residue distance maps. Subsequently, sequence feature and graph feature are concatenated by a gated pooling mechanism to form the representations of the drug and the protein. Finally, the model employs a cross-modal mutual attention module to explicitly capture bidirectional interactions between drug and protein modalities, enabling precise identification of key functional groups in drugs and critical binding residues in proteins. Extensive experiments on the Davis and KIBA benchmark datasets demonstrate that MAGNN-DTA achieves highly competitive performance compared with recent advanced methods across standard evaluation metrics. The source code and datasets are publicly available at https://github.com/HNUBIGroup/MAGNN-DTA. ## MARS-RNA: An Automated Pipeline for Correlating RNA Molecular Dynamics Trajectories with Chemical Probing Data - Source: ChemRxiv (preprints) - Date: 2026-09-08T00:00:00Z - Authors: Saeid Ekrami, Elisa Frezza - DOI: 10.26434/chemrxiv.15005800/v2 - External ID: 10.26434/chemrxiv.15005800/v2 - Keywords: Molecular Dynamics - Source URL: - Dashboard article: Abstract: Ribonucleic acid (RNA) molecules fold into complex, highly dynamic three-dimensional conformational structures that govern their biological activities. Chemical probing methods, such as SHAPE (selective 2’-hydroxyl acylation analyzed by primer extension), provide structural insights in the form of one-dimensional (1D) nucleotide reactivity profiles. However, mapping these one-dimensional profiles back to the underlying dynamic structures remains a key challenge, as traditional analyses rely on a static view based on three-dimensional (3D) high-resolution structures from X-ray crystallography or NMR, rather than accounting for the dynamic nature of the system. In this Application Note, we introduce MARS-RNA (MD Analysis for Reactivity and Structure), an automated pipeline that extracts and integrates multi-scale structural and geometric features from RNA Molecular Dynamics (MD) trajectories, combining them with secondary (2D) and 3D annotations from X3DNA-DSSR to identify features that correlate with experimental reactivity. MARS-RNA processes raw GROMACS trajectories, performs frame-by-frame structural annotation, calculates spatial centroids and physical distance fluctuations, computes Pearson/Spearman correlation coefficients against SHAPE data, and pre-processes results for machine learning pipelines. We show its value on the RNA 3′-UTR (PDB ID: 1AUD, chain A), by highlighting several single correlation with SHAPE reactivity. ## Mechanistic insights into the enantioselective activity of Fluxametamide in Tetranychus cinnabarinus. - Source: Pest management science (journals) - Date: 2026-09-08T00:00:00Z - Categories: Docking & Screening - Journal: Pest management science - DOI: 10.1002/ps.71273 - External ID: 9e8f9b68d3a164644624462df7ba86ff531f45ce - Keywords: Molecular docking, cytochrome P450, enzyme, molecular dynamics - Source URL: - Dashboard article: Abstract: BACKGROUND Fluxametamide is a chiral isoxazoline insecticide widely used for pest control. However, the enantioselective biological activity and underlying mechanisms of its enantiomers remain poorly understood. This study aimed to systematically evaluate the enantioselective acaricidal activity of fluxametamide against Tetranychus cinnabarinus. RESULTS Bioassays revealed pronounced enantioselectivity, with S-(+)-fluxametamide exhibiting significantly higher acaricidal activity than the racemate. Detoxification enzyme assays indicated significant induction of cytochrome P450 activity following sublethal exposure. Transcriptomic and RT-qPCR analyses further identified multiple P450 genes responsive to fluxametamide, among which CYP392E7 and CYP392A16 were strongly up-regulated. Molecular docking and molecular dynamics simulations demonstrated that R-(-)-fluxametamide exhibited stronger and more stable binding to these P450 enzymes than the S-(+) enantiomer. Functional assays using recombinant proteins confirmed that CYP392E7 preferentially metabolized R-(-)-fluxametamide, whereas S-(+)-fluxametamide showed lower metabolic susceptibility. Electrophysiological analyses using two-electrode voltage clamp recordings revealed that fluxametamide acts as a noncompetitive antagonist of RDL2-mediated GABA currents without direct activation, with S-(+)-fluxametamide producing the strongest inhibitory effect. CONCLUSION The enantioselective acaricidal activity of fluxametamide is governed by a dual mechanism involving preferential metabolic clearance of the R-(-)-enantiomer and stronger target-site inhibition by the S-(+) enantiomer. These findings provide mechanistic insights into the stereoselective action of fluxametamide and highlight the importance of considering enantioselectivity in the development and risk assessment of chiral acaricides. © 2026 Society of Chemical Industry. ## Mechanistically Informed Open-World Enzyme Retrieval with a Dual-Tower Graph-Sequence Model - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-08T00:00:00+00:00 - Categories: Property Prediction - Authors: Siyuan Wang, Dan Wang, Ying Ren, Xi Chen - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c02414 - Source URL: - Dashboard article: Abstract: Identifying enzymes capable of catalyzing specific chemical transformations across large sequence databases remains a major challenge in biocatalyst discovery. Conventional fingerprint-based methods capture global molecular structure but fail to represent bond-breaking and bond-forming events, limiting generalization to structurally novel reactions. We introduce a dual-track evaluation framework to distinguish true generalization from memorization, assessing retrieval on structurally isolated reactions (n = 50) within a 63,259-sequence enzyme pool. The strongest fingerprint baseline achieves R@10 = 0.020. To address this limitation, we develop GATv2-ECR, a heterogeneous dual-tower model integrating reaction-center graph encoding, a frozen ESM-2 sequence encoder, contrastive learning, and EC-aware soft reranking. GATv2-ECR achieves R@10 = 0.160 on isolated queries and R@10 = 0.308 on an out-of-distribution subset (n = 39), capturing mechanistically relevant features and supporting generalizable enzyme retrieval under open-world conditions. ## Microfluidics-Integrated Spectroscopic Technologies for Food Safety and Quality Assessment: From Complex-Matrix Processing to On-Site Decision-Making - Source: Foods (journals) - Date: 2026-09-08T00:00:00Z - Journal: Foods - DOI: 10.3390/foods15183171 - External ID: 58edfc12a24d8fdfb2bea020f83f708e153aebc2 - Keywords: s - Source URL: - Dashboard article: Abstract: Food safety and quality analysis is shifting from laboratory-based end-point testing toward faster, lower-volume and matrix-adapted on-site decision-making. Near-infrared (NIR), visible-near-infrared (Vis-NIR), hyperspectral, Raman, surface-enhanced Raman scattering (SERS), fluorescence, colorimetric and terahertz approaches, together with impedance time-series readout, provide complementary information on composition, molecular vibrations, spatial distribution, reaction outputs, or electrical responses. In real foods, however, lipids, proteins, sugars, salts, pigments, particles and native fluorescence can alter spectral baselines, mass transfer and model stability. The value of microfluidics is therefore not limited to miniaturization but lies in organizing filtration, homogenization, splitting, mixing, extraction, enrichment, reaction, and readout positions into a controllable sample-to-signal workflow. This review first distinguishes chemical hazards, biological hazards, authenticity issues, and quality changes according to target and matrix characteristics, and then compares the functional boundaries of continuous-flow, paper-based, droplet, digital-hybrid and enrichment-oriented chips. It further analyses how microfluidics affects detection time, sample and reagent consumption, sensitivity, selectivity, repeatability, portability and cross-matrix applicability through spectral interfaces, signal enhancement, labelled and label-free detection, chemometrics, and machine learning. Representative applications involving pesticides, mycotoxins, pathogens, antibiotics, heavy metals, adulterants, oxidation products, and freshness indicators in real foods are discussed within a unified chain linking chip architecture, spectral signal generation and decision models. Finally, requirements for translation are proposed in terms of standard and real samples, chip-to-chip variation, external model validation, data traceability and scalable manufacturing, providing an operational framework for the joint design of broad-spectrum spectroscopic technologies and microfluidic systems. ## Modern Psychopharmacology: Integrating Molecular Modeling, Omics Technologies, Translational Models, and Evidence-Based Clinical Studies. - Source: Psychopharmacology and Addiction Biology (journals) - Date: 2026-09-08T00:00:00Z - Categories: Cheminformatics - Journal: Psychopharmacology and Addiction Biology - DOI: 10.17816/phbn713547 - External ID: d9326b0a746374ce74f836af5bee4149a58fee1a - Keywords: cheminformatics - Source URL: - Dashboard article: Abstract: This article summarizes the outcomes of three symposia on psychopharmacology held within the framework of the XIII All-Russian Conference with International Participation “Current Issues in Preclinical and Clinical Research of Medicinal Products and Clinical Trials of Medical Devices” (St. Petersburg, April 23–24, 2026). The advancement of modern psychopharmacology requires the implementation of personalized approaches, the application of molecular modeling and computational biology methods, and the integration of pharmacogenetic, pharmacoepigenomic, and multi-omics data into drug development and efficacy assessment processes. The symposium participants identified the following priority directions for the development of contemporary psychopharmacology: (1) implementation of integrated in silico pipelines throughout all stages of drug development; (2) broader application of molecular modeling approaches for the design of innovative therapeutic agents and targeted drug delivery systems; (3) development of standardized translational models with high reproducibility and predictive value for preclinical research; and (4) integration of pharmacogenomic, pharmacoepigenomic, and multi-omics methodologies into the design and conduct of clinical studies. The major challenges facing modern psychopharmacology include: (1) further improvement of the regulatory and methodological framework governing the use of advanced digital and computational technologies, including artificial intelligence, in the development of medicinal products and medical devices; (2) standardization and validation of in silico modeling data, as well as implementation of pharmacogenetic approaches in psychiatric and addiction medicine practice; (3) development of interdisciplinary educational programs at the intersection of medicine, biology, bioinformatics, cheminformatics, and information technologies; and (4) establishment of multidisciplinary research consortia. In conclusion, the future development of psychopharmacology depends on the systematic integration of fundamental biomedical research, omics technologies, computational approaches, translational models, and high-quality clinical investigations. The implementation of such integrative projects will facilitate the development of innovative therapeutic agents and promote the advancement of personalized psychiatry and addiction medicine. ## Multi-epitope vaccine targeting SARS-CoV-2 omicron S and N proteins promotes enhanced immunity: a computational approach - Source: Frontiers in Immunology (journals) - Date: 2026-09-08T00:00:00Z - Categories: Docking & Screening - Journal: Frontiers in Immunology - DOI: 10.3389/fimmu.2026.1906485 - External ID: 57ac8ad91c67abbded728e21c74966a951c82e6d - Keywords: Molecular docking, receptor, molecular dynamics - Source URL: - Dashboard article: Abstract: The emergence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) led to the COVID-19 pandemic, which resulted in millions of deaths globally and had profound social, economic, and political consequences. Although effective vaccines and antiviral therapies have substantially reduced the global burden of COVID-19, the continued emergence of viral variants highlights the need for next-generation effective vaccine strategies capable of providing broader and more durable immune response. In this work, we provide an immunoinformatic approach for multi-epitope vaccine (MEV) design and prediction. Based on the spike (S) and nucleocapsid (N) proteins of SARS-CoV-2, immunoinformatic methods were used to identify the epitopes for B cells, cytotoxic T lymphocytes (CTL), and helper T lymphocytes (HTL). The B cell, CTL, and HTL epitopes were conjugated with flexible linkers GSG, GSGG, and a Gb-1 peptide conjugated to the C-terminal of the MEV ccandidate. The final MEV candidate exhibited favorable predicted characteristics, with a molecular weight of approximately 55.47 kDa and a length of 498 amino acid residues. Computational analyses indicated that the designed construct was antigenic, non-toxic, non-allergenic, and possessed suitable physicochemical properties and predicted solubility, supporting its potential as a vaccine candidate for further investigation. Molecular docking analysis demonstrated favorable interactions between the MEV construct and selected Toll-like receptors (TLRs), while molecular dynamics (MD) simulations suggested the stability of the vaccine-receptor complexes throughout the simulation period. Furthermore, C-ImmSim-based immune simulation predicted the induction of both humoral and cellular immune responses following the proposed immunization schedule. Collectively, these findings highlight the potential of the designed MEV construct as a computationally optimized vaccine candidate and provide a framework for future experimental evaluation. This study presents a computationally designed MEV candidate against SARS-CoV-2 by integrating immunoinformatics approaches, structural modeling, molecular docking, molecular dynamics simulations, and immune response prediction. The findings suggest that the proposed MEV construct may possess favorable immunogenic and structural properties; however, experimental validation through in vitro and in vivo studies remains essential to confirm its safety, immunogenicity, and protective efficacy. The proposed approach provides a valuable strategy for accelerating rational vaccine design and may serve as a foundation for future development of experimentally validated vaccine candidates. ## MultiRSF: A Deep Learning Approach for Predicting RNA-Small-Molecule Binding Sites Using Surface Characteristics - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-08T00:00:00+00:00 - Categories: Property Prediction, Spectra & Analytical, Free Energy & MD - Authors: Jiasai Shu, Wentao Xia, Yingjie Zheng, Yucheng Shu, Zhiyuan Zhao, Mei Feng, Yan Wang, Xiaogang Wang, Bijun Xu, Xiaojun Xu, Tingting Sun - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c02039 - Keywords: Transformer - Source URL: - Dashboard article: Abstract: Identification of RNA-small-molecule binding sites is a critical first step in RNA-targeted drug discovery. Although several machine learning methods have made progress by integrating RNA sequence, secondary structure, and 3-dimensional (3D) atomic arrangement information to identify nucleotide level binding residues, most of them neglect the molecular surface that directly contacts small-molecule compounds and are highly dependent on accurate three-dimensional structures. Here, we present MultiRSF, a multimodal deep learning framework that integrates molecular surface fingerprints, contextual sequence embeddings from a pretrained RNA language model and dot bracket secondary structure encoding to predict ligand binding sites on RNA. MultiRSF fuses these features through a hierarchical Transformer encoder and demonstrates robust predictive performance on both ligand-free RNA structures (apo RNA) and ligand-bound RNA structures (holo RNA). On independent benchmark sets TE18 and APO8, MultiRSF outperforms state-of-the-art methods, achieving precision of 0.765/0.458, recall of 0.698/0.286, and Matthews correlation coefficient (MCC) of 0.535/0.279. Case studies on a ligand-induced riboswitch conformational change and an NMR conformational ensemble further illustrate the model’s robustness to moderate RNA flexibility. In conclusion, MultiRSF provides an accurate and generalizable tool for nucleotide resolution binding sites prediction, with potential to accelerate early-stage RNA-targeted drug discovery. ## MultiRSF: A Deep Learning Approach for Predicting RNA-Small-Molecule Binding Sites Using Surface Characteristics - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-08T00:00:00Z - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c02039 - External ID: c4c9c5a9a48fee08575abdb117f572c775e96f26 - Keywords: Transformer - Source URL: - Dashboard article: Abstract: Identification of RNA-small-molecule binding sites is a critical first step in RNA-targeted drug discovery. Although several machine learning methods have made progress by integrating RNA sequence, secondary structure, and 3-dimensional (3D) atomic arrangement information to identify nucleotide level binding residues, most of them neglect the molecular surface that directly contacts small-molecule compounds and are highly dependent on accurate three-dimensional structures. Here, we present MultiRSF, a multimodal deep learning framework that integrates molecular surface fingerprints, contextual sequence embeddings from a pretrained RNA language model and dot bracket secondary structure encoding to predict ligand binding sites on RNA. MultiRSF fuses these features through a hierarchical Transformer encoder and demonstrates robust predictive performance on both ligand-free RNA structures (apo RNA) and ligand-bound RNA structures (holo RNA). On independent benchmark sets TE18 and APO8, MultiRSF outperforms state-of-the-art methods, achieving precision of 0.765/0.458, recall of 0.698/0.286, and Matthews correlation coefficient (MCC) of 0.535/0.279. Case studies on a ligand-induced riboswitch conformational change and an NMR conformational ensemble further illustrate the model’s robustness to moderate RNA flexibility. In conclusion, MultiRSF provides an accurate and generalizable tool for nucleotide resolution binding sites prediction, with potential to accelerate early-stage RNA-targeted drug discovery. ## Odorant-Binding Protein 17 Mediates the Recognition of Soybean Volatiles in Adult Riptortus pedestris - Source: Journal of Agricultural and Food Chemistry (journals) - Date: 2026-09-08T00:00:00Z - Categories: Docking & Screening - Journal: Journal of Agricultural and Food Chemistry - DOI: 10.1021/acs.jafc.6c09905 - External ID: b70edd6bb7c41804faf20558c35ee84e4e340b2e - Keywords: molecular docking - Source URL: - Dashboard article: Abstract: RpedOBP17, an odorant-binding protein highly expressed in adult Riptortus pedestris antennae, was systematically investigated using prokaryotic expression, fluorescence competitive binding, molecular docking, site-directed mutagenesis, and RNA interference. Binding assays revealed affinity to six soybean volatile compounds. Molecular docking and mutagenesis identified Ile46 and Phe129 as critical residues for ligand recognition, since alanine substitutions markedly reduced binding. RNAi-mediated knockdown significantly impaired olfactory chemotaxis, with electroantennogram recordings showing attenuated responses to (Z)-3-hexenyl acetate, (E)-2-hexenyl acetate, and 4-ethylbenzaldehyde in both sexes, along with sex-specific changes to 2-methyl-1-butanol (females) and (Z)-3-hexenyl propionate (males). Y-tube olfactometer tests further confirmed decreased attraction to (E)-2-hexenyl acetate. These results elucidate the molecular interaction mechanism and provide a theoretical foundation for developing novel olfactory disruption-based control agents targeting RpedOBP17. ## On the Number of Metric Bases of Tyrosine Kinase Inhibitors - Source: Discrete Mathematics, Algorithms and Applications (journals) - Date: 2026-09-08T00:00:00Z - Categories: Cheminformatics - Journal: Discrete Mathematics, Algorithms and Applications - DOI: 10.1142/s179383092650093x - External ID: a77f7c7633a59e16f64831c5f4847c612b2961be - Keywords: cheminformatics, Kinase - Source URL: - Dashboard article: Abstract: Tyrosine kinase inhibitors (TKIs) are a type of targeted cancer drugs that work by blocking particular pathways that promote angiogenesis and tumour growth. Although the biochemical mechanisms of TKIs have been well studied, further understanding of their complexity, classification, and possible combinatorial behaviour can be gained by structural graph-theoretic research. In this study, we determine the metric dimension and all possible metric bases of molecular graphs corresponding to selected TKIs such as Sunitinib, Sorafenib, Axitinib, Regorafenib, Cabozantinib, Pazopanib and Lenvatinib. By identifying the metric bases of these molecular structures, we may investigate how structural distinguishability can be used in cheminformatics methods to drug repurposing, molecular categorization, and combination therapy modelling. While the real-world chemical information by considering atoms and bonds as discrete vertices and edges can be simplified by molecular graphs, the analysis of them remains useful in computational drug discovery. This work enables more research at the intersection of discrete mathematics and biomedical science and helps in mathematical characterization of anti-cancer drugs. ## Paeoniflorin prevents acute graft-versus-host disease while preserving graft-versus-tumor effects - Source: Frontiers in Immunology (journals) - Date: 2026-09-08T00:00:00Z - Categories: Docking & Screening - Journal: Frontiers in Immunology - DOI: 10.3389/fimmu.2026.1937379 - External ID: 9e2b7b7b8ec6641a62b5151f17680a55cb1ac382 - Keywords: molecular docking - Source URL: - Dashboard article: Abstract: Acute graft-versus-host disease (aGVHD) remains a major complication after allogeneic hematopoietic stem cell transplantation (allo-HSCT). This study investigated the preventive effects of paeoniflorin (PF), a natural monoterpene glycoside, on aGVHD. Candidate herb-derived compounds against aGVHD were screened using SymMap, HERB, ADMETlab, and InflamNat, followed by molecular docking. The preventive effects of PF were evaluated in major histocompatibility complex (MHC)-mismatched murine aGVHD models. An entire-process paeoniflorin (EP-PF) regimen, involving donor pretreatment before transplantation and recipient treatment after transplantation, was proposed for enhancing the preventive effect. Survival, body weight, clinical score, hematopoietic reconstitution, donor T cell infiltration, cytokine and chemokine levels, gut injury, apoptosis, gut microbiota, transcriptomic changes, and graft-versus-tumor (GVT) activity were assessed. PF was identified as a promising multi-target compound for aGVHD prevention. PF reduced clinical severity, body weight loss, and mortality, while improving hematopoietic recovery and increasing hematopoietic stem/progenitor cell (HSPC) numbers. EP-PF regimen showed stronger protection than recipient-only PF administration. Mechanistically, EP-PF reduced intestinal donor cluster of differentiation 4 (CD4) positive and cluster of differentiation 8 (CD8) positive T cell infiltration, partially restored CD4 + /CD8 + T cell balance, and reversed inflammatory transcriptional programs in donor T cells. Moreover, EP-PF balanced serum levels of cytokines and chemokines, preserved gut barrier integrity, reduced intestinal apoptosis, and partially corrected gut microbiota dysbiosis. Importantly, EP-PF preserved donor cell-mediated tumor control in a green fluorescent protein (GFP)-labeled A20 GVT model. PF also directly inhibited the proliferation of hematologic tumor cells and promoted their apoptosis in vitro and in vivo . PF prevents experimental aGVHD through coordinated regulation of donor T cell responses, inflammatory mediators, hematopoietic reconstitution, intestinal barrier injury, apoptosis, and gut microbiota, while preserving GVT activity. These findings support PF as a potential prophylactic candidate for allo-HSCT. ## PARAM-DOCK: An Integrated Framework for Parallel Multi-Protein–Multi-Ligand Docking and Scoring for Scalable Structure-Based Drug Discovery - Source: ChemRxiv (preprints) - Date: 2026-09-08T00:00:00Z - Categories: Property Prediction, Docking & Screening, ADMET & Safety - Authors: Samarth Kittad, Mallikarjunachari Uppuladinne, Aneesh Kotipalli, Shruti Koulgi, Vinod Jani, Uddhavesh Sonavane - DOI: 10.26434/chemrxiv.15008491/v1 - External ID: 10.26434/chemrxiv.15008491/v1 - Keywords: Molecular docking, ADMET prediction, receptor, ADMET - Source URL: - Dashboard article: Abstract: Molecular docking is a central component of structure-based drug discovery, but practical docking studies often require several independent programs for receptor and ligand preparation, binding-site definition, docking, visualisation, interaction analysis, and downstream compound assessment. This fragmentation becomes particularly limiting when multiple protein targets and multiple ligands must be evaluated in the same study. Here, we present PARAM-DOCK, an open-source molecular docking engine and integrated web workflow implemented in Python with Numba-compiled numerical kernels. PARAM-DOCK supports parallel multi-protein-multi-ligand docking and automates major pre-processing and post-processing steps within a single platform. The docking engine combines a seven-term empirical scoring function with an iterated local search strategy in which Monte Carlo perturbations are followed by bounded quasi-Newton minimisation using analytical gradients. Final pose selection can additionally incorporate binding-basin population and pocket-complementarity descriptors when they are informative. The web platform integrates molecular preparation, docking setup, pose ranking, 2D and 3D protein-ligand interaction analysis, comparative score and interaction heatmaps, molecular visualisation, and ADMET prediction. PARAM-DOCK was benchmarked against Vina using matched receptor and ligand PDBQT files, identical search boxes, symmetry-corrected RMSD evaluation, and identical hardware. On the 285-complex CASF-2016 core set, PARAM-DOCK achieved a Pearson correlation of 0.631 for scoring power and a mean within-target Spearman correlation of 0.579, compared with 0.604 and 0.528 for Vina. In de novo redocking of 255 complexes, PARAM-DOCK placed the top-ranked pose within 2.0 Å in 68.6% of cases and recovered a near-native pose among nine outputs in 84.3%, compared with 74.1% and 93.7% for Vina. Conditional on generating a near-native pose, PARAM-DOCK ranked it first in 81.4% of complexes versus 79.1% for Vina. These results indicate competitive scoring and ranking performance, while conformational sampling remains the main area for further improvement. ## Physics-informed machine learning framework integrating solid solution strengthening theory for accelerated hardness prediction in high-entropy alloys - Source: Journal of Materials Informatics (journals) - Date: 2026-09-08T00:00:00Z - Journal: Journal of Materials Informatics - DOI: 10.20517/jmi.2026.16 - External ID: b3cc00eafaf38cb8db10080b59b7ee959943a78c - Keywords: virtual screening - Source URL: - Dashboard article: Abstract: High-entropy alloys (HEAs) exhibit exceptional stability in extreme environments, yet their expansive design space presents a “curse of dimensionality” for traditional discovery methods. While machine learning (ML) offers a data-driven paradigm for material screening, the scarcity of experimental data often results in overfitting and limited physical interpretability. To address these challenges, this study proposes a hybrid physics-informed machine learning (Hybrid PIML) framework for accelerated hardness prediction. By integrating classical solid solution strengthening theory with a residual learning artificial neural network (ANN), the model explicitly embeds the physical coupling of shear modulus and lattice distortion (G·δr2/3) as prior knowledge. This approach ensures predictions adhere to metallurgical principles while significantly outperforming benchmark algorithms, achieving a coefficient of determination (R2) of 0.976 and reducing the root mean square error (RMSE) by approximately 43%. SHapley Additive exPlanations (SHAP) analysis confirms that physics-enhanced features dominate the decision-making process, validating the model’s internalization of strengthening mechanisms. Furthermore, the research elucidates a phase-dependent non-linear correlation between hardness and yield strength, correcting the failure of the classical Tabor formula in work-hardening face-centered cubic (FCC) alloys. Finally, a high-throughput virtual screening funnel based on this framework successfully identified optimized non-equiatomic candidates within the refractory Co-Cr-Ti-Mo-W system. This work establishes a precise, physically consistent pathway for inverse material design under data-constrained conditions. ## Precision Drug Discovery in the Era of Artificial Intelligence: A Critical Review. - Source: Annual review of pharmacology and toxicology (journals) - Date: 2026-09-08T00:00:00Z - Categories: ADMET & Safety - Journal: Annual review of pharmacology and toxicology - DOI: 10.1146/annurev-pharmtox-060325-010428 - External ID: ab9ac9d7f0d5053ce0b3a6b54f86fbf194efa335 - Keywords: virtual screening - Source URL: - Dashboard article: Abstract: Traditional drug development suffers from high costs, low success rates, and patient response variability. Precision drug discovery seeks to overcome these limitations by targeting specific genetic and molecular mechanisms but faces challenges in integrating cross-scale, multimodal biomedical data. Recent advances in artificial intelligence (AI), especially deep learning, provide powerful tools to navigate these complexities. This review surveys representative AI methods across four core stages of precision drug discovery: (a) target identification and validation using omics-, drug-, and structure-based approaches; (b) structure- and sequence-guided virtual screening of active compounds; (c) individualized drug response prediction integrating cell-line, single-cell, and multimodal data; and (d) AI-enabled toxicology and safety modeling to anticipate adverse liabilities and improve translational success. We further highlight a paradigm shift toward autonomous scientific agents capable of causal reasoning and end-to-end experimental guidance. Finally, we discuss persistent challenges, including data bias, limited interpretability, and in silico-to-wet lab translation. ## Regulatory Network Unveils the Molecular Architecture Governing Seed Vigor in Watermelon (Citrullus lanatus). - Source: The Plant cell (journals) - Date: 2026-09-08T00:00:00Z - Categories: Docking & Screening - Journal: The Plant cell - DOI: 10.1093/plcell/koag263 - External ID: af6a0bce61e7bc77969fc4bc129c5f630acd0d21 - Keywords: molecular docking - Source URL: - Dashboard article: Abstract: Seed vigor, a determinant of agricultural productivity, diminishes during ageing but can be restored through priming. We integrated multi-omics approaches to dissect molecular mechanisms governing watermelon seed vigor under controlled deterioration and priming. Controlled deterioration (CD) progressively impaired germination speed (t50) and uniformity (AUC) before reducing maximum germination (Gmax), whereas priming reversed these effects even under ageing conditions. Transcriptome analyses revealed that ageing upregulated oxidative-stress genes (e.g., cytochrome c oxidase) and suppressed translation machinery. Priming activated stress-response pathways and ribosomal proteins through DOF/ERF-mediated transcriptional reprogramming via cis-element recognition, and molecular docking and yeast one-hybrid assays confirmed this interaction. Two biomarkers, Cla97C04G070560 (cytochrome c oxidase) and Cla97C03G062270 (ribosomal protein), were identified as quantitative predictors of germination decline. Machine-learning-based proteomic models and a regulatory network, SeedWatermelonNet (SWN), were constructed. These analyses uncovered hub genes (e.g., OHCU\_decarbox, HSPs, LIM, NADPH-dependent aldehyde reductase 1, RPL32) coordinating uric acid metabolism, stress responses, energy metabolism, and translation. Experimental validation confirmed the regulatory roles of these candidates, including physical interactions mediating seed-vigor transduction cascades and their predicted function as direct regulators of seed vigor via SWN. Maternal alleles dominated transcriptional regulation of stored mRNAs, with ageing reducing maternal-specific gene expression. Promoter SNPs in these genes implicated BPC and DOF transcription factors in maternal bias. These findings establish molecular thresholds for vigor loss, validate priming as a resilience strategy, and provide biomarkers for seed quality assessment. Integrating transcriptomic, proteomic, and allele-specific insights advances precision breeding and storage practices, offering targets to enhance seed resilience against climate-driven agricultural challenges. ## Reinforcement Learning-Assisted Quantum Simulation of Many-Body Excited States and Real-Time Dynamics - Source: Journal of Chemical Theory and Computation (journals) - Date: 2026-09-08T00:00:00+00:00 - Authors: Jiaji Zhang, Lipeng Chen, Carlos L. Benavides-Riveros - Journal: Journal of Chemical Theory and Computation - DOI: 10.1021/acs.jctc.6c00988 - Keywords: Reinforcement Learning, RL - Source URL: - Dashboard article: Abstract: The computation of electronic excited states and real-time quantum dynamics of many-Fermion systems is among the most promising applications of near-term quantum computing. In this work, we generalize the reinforcement learning contracted quantum eigensolver (RL-CQE), previously developed for ground-state problems, to electronic excited states and real-time quantum dynamics, in which a deep Q-network agent adaptively selects the two-body operators at each iteration, yielding more compact ansätze and improved robustness with respect to critical hyperparameters. A key feature of the algorithm is a scalable state representation based on the ACSE residuals, whose dimension grows with the one-particle basis but remains independent of the number of targeted excited states. We also show the equivalence of sign-free qubit operators in the excited-state and time-evolution settings, extending a result previously established for ground-state problems. Our RL-CQE for time evolution derives from a constant-scaling ansatz that represents the wave function with a fixed number of unitary transformations independent of simulation time, enabled by the shared unitary structure of the purified ensemble treatment of excited states. Benchmarks on chemical systems demonstrate chemical accuracy with minimal operator counts across a range of bond lengths. ## Reliable Structure-Processing-Performance Learning in Organic Photovoltaics: Chemical Generalization, Processing-Aware Prediction and Uncertainty under Domain Shift - Source: ChemRxiv (preprints) - Date: 2026-09-08T00:00:00Z - Categories: Cheminformatics - Authors: DENNIS OBINNA ORJI - DOI: 10.26434/chemrxiv.15008466/v2 - External ID: 10.26434/chemrxiv.15008466/v2 - Keywords: RDKit - Source URL: - Dashboard article: Abstract: Machine-learning models for organic photovoltaics (OPVs) can achieve strong performance under random train-test splits, but such evaluation may overstate prospective generalization when literaturederived datasets contain repeated publications and material systems. Here, two complementary OPV benchmarks were constructed to separate interpolation from chemical extrapolation: a provenanceclean OPV-DB cohort containing 21,590 experimentally reported devices and a processing-rich Wen-Zhang-Ma cohort containing 994 unique physical processing conditions. Random, publication-heldout, donor-acceptor (D:A)-pair-held-out, acceptor-held-out, and donor-held-out validation regimes were compared, together with structure-only, processing-only, and combined structure-processing representations. Extra Trees performance on OPV-DB declined from R2 = 0.813 under random validation to R2 = 0.362 under donor holdout. In the processing-rich benchmark, adding processing information reduced MAE by 19.5% under random validation but by only 1.0-2.4% under chemically grouped validation. This effect was strongly associated with material-system familiarity: processing reduced MAE by 27.1% for test records whose D:A pair was represented in training, compared with 1.8% for completely unseen pairs. Ensemble dispersion was useful for ranking errors under random validation (Spearman rho = 0.462) but lost reliability under donor holdout (rho = 0.047), where the most uncertain 20% of predictions captured only 16.1% of the largest errors. A cross-dataset transfer audit further showed that a common unmodified RDKit representation covered 73.54% of the processing benchmark and only 38.02% of records with PCE >= 15%, creating substantial chemicalselection bias. These results show that validation design is central to claims of OPV model reliability, that processing information primarily refines predictions within familiar chemistry, and that both uncertainty estimation and transfer benchmarking require explicit domain-shift and representation audits. ## Robust Multi-Objective Reinforcement Learning Improves Joint Predicted-Potency and Drug-Likeness Yield in EGFR Molecule Generation - Source: ChemRxiv (preprints) - Date: 2026-09-08T00:00:00Z - Categories: Design de novo - Authors: Yutong Guo, Xiao Huang - DOI: 10.26434/chemrxiv.15008490/v1 - External ID: 10.26434/chemrxiv.15008490/v1 - Keywords: Molecule Generation, Drug Likeness, ChemBERTa, ChEMBL, Reinforcement Learning, RL, GNN, IC50, pIC50, receptor - Source URL: - Dashboard article: Abstract: Optimizing a molecular generator against a single activity-model estimate can reward structures for which that estimate is unreliable. We extended our published ChemBERTa–reinforcement-learning (RL) workflow for human epidermal growth factor receptor (EGFR) with scaffold-separated activity modeling, empirically calibrated lower scores, bounded chemistry rewards, and matched-seed evaluation. From ChEMBL 37, we curated 6,476 unique structures with Reference/unspecified biochemical EGFR IC50 measurements; 4,491, 680, and 1,305 structures were assigned to scaffold-disjoint training, calibration, and test sets. The random-forest reward model achieved a test mean absolute error of 0.610 pIC50 units and R2 = 0.615. Uniform or potency-enriched masked-language-model continuation was crossed with three RL rewards, with three seeds and 600 generation attempts per combination. The locked LCB-versus-mean-only comparison changed GNN potency-screen yield by +0.28 percentage points (95% descriptive t interval,-1.10 to +1.66) across three seed-cluster means. In the pipeline-level comparison with the published-form baseline, Robust-LCB increased the GNN joint-screen yield in all six matched blocks; averaged across the two MLM arms, the yield rose from 0.72% to 1.22%. Two structures passed the full RF–GNN consensus funnel. These results support a pipeline-level increase in joint predicted-potency and drug-like yield, while the incremental contribution of the lower-score term alone remains uncertain. These comparisons describe computational variability; activity estimates for generated molecules require biochemical testing. ## Scaffold-Controlled Evaluation of Molecular Representations for Antimalarial Activity Prediction - Source: ChemRxiv (preprints) - Date: 2026-09-08T00:00:00Z - Authors: Jean-Pierre TCHAPET NJAFA, Penabei SAMAFOU, Fon Wilfred MBACHAM, Serge Guy NANA ENGO, Myke Vital SAO TEMGOUA - DOI: 10.26434/chemrxiv.15008517/v1 - External ID: 10.26434/chemrxiv.15008517/v1 - Keywords: Activity Prediction, Molecular Representations - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Substrate α-1,4-Glucan Structural Organization Determines EGCG Inhibition of α-Amylase through an Enzyme-Substrate-Inhibitor Tug-of-War - Source: Journal of Agricultural and Food Chemistry (journals) - Date: 2026-09-08T00:00:00Z - Categories: Docking & Screening - Journal: Journal of Agricultural and Food Chemistry - DOI: 10.1021/acs.jafc.6c11481 - External ID: 8784712fcf835e5d3a85f7c89a267a589c028b77 - Keywords: molecular docking, Enzyme, molecular dynamics - Source URL: - Dashboard article: Abstract: α-Amylase inhibitors often exhibit markedly inconsistent activities across substrate systems, yet the molecular basis remains poorly understood. Here, using epigallocatechin gallate (EGCG) as a model inhibitor, we compared corn starch, a continuous α-1,4-glucan, with the small, structurally defined chromogenic substrate GalG2CNP, demonstrating that substrate structural organization, rather than the inhibitor itself, governs apparent inhibitory outcomes. Their distinct enzyme affinities produced opposing responses: low-affinity GalG2CNP was readily inhibited by EGCG, whereas high-affinity starch resisted inhibition even with excess inhibitor. Multimodal biophysics, molecular docking, and molecular dynamics simulations confirmed that EGCG occupies the α-amylase active pocket without substantially perturbing enzyme structure. Ternary system analyses further revealed that high-affinity starch displaced prebound EGCG, whereas GalG2CNP induced only partial re-equilibration. We therefore propose a substrate-affinity-driven tug-of-war model and a two-tier screening strategy, using artificial substrates for primary screening and starch for physiological verification, providing a mechanistic framework for more reliably evaluating candidate dietary α-amylase inhibitors. ## SynOmega: Simplifying Retrosynthesis for Efficient Synthesizability Scoring - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-08T00:00:00+00:00 - Categories: Cheminformatics, Reaction Informatics - Authors: Baicheng Zhang, Guoqing Zhang, Jun Jiang, Yi Luo - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c02529 - Keywords: AiZynthFinder, ChEMBL, Retrosynthesis - Source URL: - Dashboard article: Abstract: Retrosynthesis-based synthesizability scoring triages molecules from generative design but is expensive: every score requires a multi-step search. We present SynOmega, an open-source toolkit that couples a single-step template model, an AND–OR route search, and a route-based synthesizability score (SynScore). Its single-step model can be restricted, at the reaction-template level, to simplifying disconnections that split the target into smaller precursors. On 1000 ChEMBL drug molecules this yields two findings. (1) The simplifying constraint cuts node expansions by about 30% on jointly solved targets and lowers the median search time by about a third. This reduction in search effort is the robust, budget-independent result; the small accompanying rise in solved rate is a secondary effect between the two separately trained models, not the isolated result of toggling one model’s action space. (2) As a complete system, under matched search depth, width and iteration budget, SynOmega reaches about 1.8× the solved rate of the open-source planner AiZynthFinder while searching about 13× faster. SynOmega thus offers a cheap, data-level action-space constraint that makes route-based synthesizability scoring more efficient without sacrificing solvability. ## SynOmega: Simplifying Retrosynthesis for Efficient Synthesizability Scoring - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-08T00:00:00Z - Categories: Reaction Informatics - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c02529 - External ID: 905b8bebd010d0b02daa3843dcb82948a88b82ec - Keywords: AiZynthFinder, ChEMBL, Retrosynthesis - Source URL: - Dashboard article: Abstract: Retrosynthesis-based synthesizability scoring triages molecules from generative design but is expensive: every score requires a multi-step search. We present SynOmega, an open-source toolkit that couples a single-step template model, an AND–OR route search, and a route-based synthesizability score (SynScore). Its single-step model can be restricted, at the reaction-template level, to simplifying disconnections that split the target into smaller precursors. On 1000 ChEMBL drug molecules this yields two findings. (1) The simplifying constraint cuts node expansions by about 30% on jointly solved targets and lowers the median search time by about a third. This reduction in search effort is the robust, budget-independent result; the small accompanying rise in solved rate is a secondary effect between the two separately trained models, not the isolated result of toggling one model’s action space. (2) As a complete system, under matched search depth, width and iteration budget, SynOmega reaches about 1.8× the solved rate of the open-source planner AiZynthFinder while searching about 13× faster. SynOmega thus offers a cheap, data-level action-space constraint that makes route-based synthesizability scoring more efficient without sacrificing solvability. ## Synthesis and Antidiabetic Evaluation of Novel 2,4-Thiazolidinedione Derivatives Targeting Key Carbohydrate-Digesting Enzymes - Source: Molecules (journals) - Date: 2026-09-08T00:00:00Z - Categories: Docking & Screening, ADMET & Safety - Journal: Molecules - DOI: 10.3390/molecules31183160 - External ID: 14cc7a199cc8ea49cd9f7cdc2e7d6e787138feb2 - Keywords: Molecular docking, drug likeness, Lipinski, IC50, ADME - Source URL: - Dashboard article: Abstract: Diabetes mellitus is a long-term metabolic disease associated with elevated glucose levels in blood and still constitutes one of the major public health issues worldwide. Inhibition of carbohydrate-digesting enzymes like α-amylase and α-glucosidase has been found to be effective in controlling postprandial hyperglycemia. The current study focused on designing, synthesis, characterization, and evaluation of novel 2,4-thiazolidinedione derivatives (D1–D5) as potent antidiabetic drugs utilizing combined in silico, in vitro, and in vivo techniques. Results from drug-likeness and ADME analyses indicated that all synthesized derivatives met Lipinski’s rule of five and had desirable pharmacokinetics properties along with reduced toxicity. Molecular docking against maltase-glucoamylase (human; PDB ID: 3TOP) protein showed good binding affinities of both D1 and D5 derivatives (−7.74 and −7.40 kcal/mol respectively) due to stable interactions with catalytic residues of enzymes. Inhibition of enzymes in vitro showed that D1 and D5 had the highest inhibitory activities of all synthesized derivatives, with IC50 of 33.86 ± 2.1 and 37.55 ± 1.7 μM against α-amylase and 29.81 ± 3.2 and 32.43 ± 1.2 μM against α-glucosidase, respectively. Cytocompatibility tests on L6 myoblast cells proved that the lead compounds were well tolerated. In addition, studies in a model of Drosophila melanogaster induced by a high-sugar diet revealed a significant decrease in the level of glucose concentration depending on the dose, especially for D1 and D5, indicating their antihyperglycemic activity in vivo. Thus, these data confirm that D1 and D5 can be regarded as promising lead compounds for the development of new antidiabetics acting via inhibition of carbohydrate-metabolizing enzymes. ## Synthesis of Quinazoline Derivatives and Mechanistic Approaches in Lung and Breast Cancers - Source: Molecules (journals) - Date: 2026-09-08T00:00:00Z - Categories: Docking & Screening - Journal: Molecules - DOI: 10.3390/molecules31183163 - External ID: fd734480902091f142884cf2cb73bfad70001d28 - Keywords: molecular docking, IC50, molecular dynamics - Source URL: - Dashboard article: Abstract: The quinazoline/quinazolinone ring is known as a unique scaffold, and its derivatives possess a broad biological activity profile, including antibacterial, antifungal, anticonvulsant, anti-inflammatory, anti-HIV, and analgesic activity, primarily focusing on anticancer activity. In this study, the synthesis of 2-\[\[4-oxo-3-(substituted phenyl)-3,4-dihydro-(substituted quinazolin-2-yl)\]thio\]-N′-(aryl/heteroaryl methylene)acetohydrazide (4a–4x) derivatives and their potential anticancer activities were investigated on the lung cancer A549 cell line, the breast cancer MCF-7 cell line, and healthy fibroblast L929 cell line. Compounds 4c, 4i, 4m, and 4u were identified as the most cytotoxic and selective molecules on the A549 cell line (IC50: 44.75–78.13 µM), while 4i, 4l, 4m, and 4u were identified as the most cytotoxic and selective molecules on the MCF-7 cell line (IC50: 14.43–29.39 µM). The mechanisms of action of their anticancer activities were examined and studied. It was determined that these compounds induced strong apoptosis and significantly activated caspase-3 activation in both cell types and that they interrupted the cell cycle in the pre-G (sub G0) phase. Compounds 4m and 4u exhibited EGFR inhibition (IC50: 4.80 µM, IC50: 5.40 µM, respectively) at a level similar to the standard drug gefitinib (IC50: 1.86 ± 0.43 µM). Based on the results of the biological activity assays, molecular docking, and molecular dynamics simulation studies, the 4-quinazolinone–acetyl hydrazone scaffold can be considered a promising structural framework with potential anticancer activity. More specifically, the findings of this study indicate that the acetyl moiety may function as an important pharmacophoric group, while the trisubstituted quinazolinone core may represent a favorable structural feature for caspase-3 activation. However, the same structural framework appears to be less favorable for EGFR inhibition, possibly due to steric constraints within the EGFR binding pockets. ## Synthesis, Herbicidal Activity Evaluation, and Molecular Docking of Novel Acylthioureas as AHAS Inhibitors - Source: Molecules (journals) - Date: 2026-09-08T00:00:00Z - Categories: Docking & Screening - Journal: Molecules - DOI: 10.3390/molecules31183162 - External ID: 568791ed96152a265a0b59bde20cb28457c7acd1 - Keywords: Molecular Docking, binding affinity, bioisosteric replacement, enzyme - Source URL: - Dashboard article: Abstract: Acetohydroxyacid synthase (AHAS, EC 2.2.1.6) is a core enzymatic target in agrochemical research for herbicide development and has been widely investigated in recent decades. To develop novel AHAS-targeted herbicides, twenty-three acylthiourea derivatives were synthesized via fragment recombination and bioisosteric replacement strategies in this work. All target compounds were fully characterized by elemental analysis, mass spectrometry, FTIR spectroscopy, and 1H NMR spectroscopy. Dose–response trends from the Petri dish assay at 1, 10, and 100 mg L−1 show that root-growth inhibition increased synchronously with concentration. Preliminary bioassays across gradient concentrations (1, 10, 100 mg L−1) revealed dose-dependent growth-inhibitory effects of several derivatives against the monocot weed Digitaria adscendens and dicot weed Amaranthus retroflexus. At 100 mg/L pre-emergence treatment, compounds 4v (76.48 ± 1.47%), 4m (69.34 ± 1.62%), and 4l (67.77 ± 1.87%) exhibited the strongest inhibitory activity, comparable to or exceeding bensulfuron-methyl (70.41 ± 1.21%). All synthesized acylthiourea derivatives exhibited less than 20% growth inhibition toward wheat and soybean, demonstrating acceptable crop selectivity. In vivo enzymatic assays at 100 mg L−1 showed that 4l, 4m, and 4v achieved AHAS inhibition rates of 38.25 ± 1.81%, 35.74 ± 1.35%, and 38.75 ± 1.93%, comparable to or marginally exceeding that of bensulfuron-methyl (35.64 ± 1.40%). Molecular docking simulations yielded binding energies of −6.67 kcal mol−1 (4l), −6.29 kcal mol−1 (4m), and −6.90 kcal mol−1 (4v), all more favorable than −5.86 kcal mol−1 calculated for bensulfuron-methyl, indicating stronger target-enzyme binding affinity for these three compounds. This research suggests that these acylthiourea derivatives may serve as preliminary lead scaffolds for developing novel AHAS inhibitors via subsequent structural derivatization, pending further dose–response, mechanistic, and field-efficacy validation. ## Targeting the 3C Protease of Hepatitis A Virus Subgenotype IB: Virtual Screening and Identification of Potent Lead Candidates - Source: Microorganisms (journals) - Date: 2026-09-08T00:00:00Z - Journal: Microorganisms - DOI: 10.3390/microorganisms14091987 - External ID: 1bc1032dbec2d235978f21d28d8f5b54dda94433 - Keywords: AutoDock Vina, Virtual Screening, In silico screening - Source URL: - Dashboard article: Abstract: Hepatitis A virus (HAV) infection remains a global public health concern in both developing and developed countries. In the present study, we identified anti-HAV drugs, using AutoDock Vina Modeling software, and evaluated the compounds in vitro. Following cytotoxicity for Huh7 cells, 5 out of 10 compounds were selected. We evaluated effective HAV 3C protease inhibitors with activity against both HAV genotype IB HM175/18f and HAV genotype IIIA HA11-1299-infected human hepatoma cells. Among the five compounds, we identified only one (KCN-A-12), which had an inhibitory effect on both HAV genotype IB HM175/18f and HAV genotype IIIA HA11-1299 replication in human hepatoma Huh7 cells. KCN-A-12 has more effective inhibitory effects on the replication of HAV genotype IB HM175/18f than that of HAV genotype IIIA HA11-1299. This difference may be attributable to the fact that our discovery system depends on crystal structures from HAV 3C protease based on the HAV genotype IB HM175 strain. In conclusion, we observed that KCN-A-12 was able to inhibit HAV replication. In silico screening for HAV 3C protease inhibitors may be useful for further discovery of anti-HAV drugs. ## The Open-Shell Organic Systems Database: Diverse CCSD(T)/CBS Reaction Energies for Assessing Density Functional Theory and Machine Learning Methods - Source: The Journal of Physical Chemistry A (journals) - Date: 2026-09-08T00:00:00Z - Journal: The Journal of Physical Chemistry A - DOI: 10.1021/acs.jpca.6c03915 - External ID: b95bafe35d4f261f7a86ef1fd0fdc374a7872035 - Keywords: ANI 1x, Density Functional Theory, DFT, covers the chemical space - Source URL: - Dashboard article: Abstract: We present the Open-Shell Organic Systems (OSOS) database, a diverse benchmark set comprising 9934 atomization, isomerization, and H-abstraction energies computed at the CCSD(T)/CBS limit via W1-F12 theory. Historically, large-scale data sets (e.g., QM9 and ANI-1x) have been heavily biased toward closed-shell species. The OSOS database addresses this gap by providing a diverse and accurate set of reaction energies dedicated exclusively to organic doublet radical systems for benchmarking and training purposes. The data set covers the chemical space spanned by localized and delocalized organic radicals containing up to seven non-hydrogen atoms (C, N, O, and F). The database consists of 3360 radical total atomization energies, 3360 H-abstractions, 2292 H-shifts, and 922 constitutional isomerizations. We use this database to assess the performance of density functional theory (DFT). We find that most density functionals systematically overestimate radical atomization energies; therefore, the inclusion of dispersion corrections generally leads to a deterioration in performance. Furthermore, delocalized radicals present a significantly greater challenge for most DFT methods than localized species, leading to larger overestimations of their atomization energies. For localized radicals, errors increase in the order: O-centered < C-centered < N-centered. Remarkably, the deep-learning meta-GGA Skala functional demonstrates exceptional accuracy, significantly outperforming conventional functionals across all rungs of Jacob’s Ladder and consistently achieving mean absolute deviations below 1 kcal mol–1 for the abstraction, shift, and isomerization reactions. The OSOS database establishes a rigorous standard to guide the development and validation of next-generation density functionals and machine-learning potentials for doublet radicals. ## Therapeutic Potential of Selected Flavonoids as Natural Antidiabetic Agents - Source: Nutrients (journals) - Date: 2026-09-08T00:00:00Z - Categories: Docking & Screening - Journal: Nutrients - DOI: 10.3390/nu18182944 - External ID: 4434d77017622151954574ec7ba06ce76145ea12 - Keywords: molecular docking, enzyme - Source URL: - Dashboard article: Abstract: Type 2 diabetes mellitus (T2DM) remains a major clinical challenge, with postprandial hyperglycemia (PPG) playing a key role in the development of vascular complications. Current pharmacotherapies often do not sufficiently control PPG, highlighting the need for complementary strategies. One effective approach is the inhibition of carbohydrate digesting enzymes, α-amylase and α-glucosidase, which delays glucose absorption and reduces PPG excursions. Flavonoids, a diverse class of plant-derived polyphenols, exhibit multiple biological activities relevant to T2DM, including enzyme inhibition, antioxidant, and anti-inflammatory effects. This review summarizes the mechanisms by which flavonoids modulate glucose metabolism, with particular emphasis on their interactions with α-amylase and α-glucosidase. Structure–activity relationships are discussed, focusing on the influence of hydroxylation patterns, conjugation, and molecular planarity on inhibitory potency and selectivity. Selected flavonoids—chrysin, apigenin, luteolin, and quercetin—are comparatively analyzed in terms of enzyme inhibition profiles, binding mechanisms, and selectivity. Evidence from in vitro studies, enzyme kinetics, and molecular docking highlights their differential activity, particularly the preferential inhibition of α-glucosidase over α-amylase, which may reduce gastrointestinal side effects. Overall, flavonoids represent promising modulators of PPG. However, their clinical relevance is limited by bioavailability and variability of experimental data, warranting further investigation. ## UbC4 from Leishmania is a Druggable E2 Ubiquitin Conjugating Enzyme: Structural Basis and Fragment Hits for Future E2-Recruiting PROTAC Development - Source: ACS Omega (journals) - Date: 2026-09-08T00:00:00+00:00 - Categories: Docking & Screening - Authors: Cécile Exertier, Lorenzo Antonelli, Anastasia Liuzzi, Marcus Ruffa, Vittorio Brufani, Gianni Colotti, Annarita Fiorillo, Andrea Ilari - Journal: ACS Omega - DOI: 10.1021/acsomega.6c05158 - Keywords: molecular docking, Enzyme - Source URL: - Dashboard article: Abstract: Leishmaniasis is a neglected disease that affects around two million people every year. Current treatments are often highly toxic or prone to resistance, underscoring the urgent need for new therapeutic strategies. PROTACs may offer a promising alternative, as they can potentially mitigate both toxicity and resistance. However, very little is known about the ubiquitin–proteasome system (UPS) in Leishmania. Notably, only two E3 ligases containing a CULT domain have been identified so far, and none carrying a von Hippel–Lindau (VHL) domain─the classical E3 ligase used by clinically advanced PROTACs. In this work, we take an important step toward understanding the UPS in Leishmania, and we propose that in this organism the UbC4 E2 enzyme, rather than an E3, may be directly exploited to develop a PROTAC able to engage a protein of interest. Here, we report the biochemical and structural characterization of the Leishmania major ubiquitin-conjugating enzyme 4 (UbC4). Through a fragment screening campaign, we identified 10 fragments binding to distinct cavities on UbC4. Among these, five interact with the same noncatalytic pocket that is poorly conserved in humans, while one fragment binds near the catalytic cysteine. Using DeepFrag predictions and molecular docking, we explored fragment elongation strategies to enhance affinity for their respective binding sites, with the goal of guiding the development of E2-recruiting PROTACs or UPS inhibitors for the treatment of leishmaniasis. ## Unveiling the anti-diarrheal potential of Persicaria vivipara: phytochemical characterization and opioid receptor-targeted in vivo and in silico studies - Source: RSC Advances (journals) - Date: 2026-09-08T00:00:00Z - Categories: Docking & Screening, ADMET & Safety - Journal: RSC Advances - DOI: 10.1039/d6ra06848k - External ID: 9bfbb50010ed1799e0a414fb5f61127d791753e2 - Keywords: molecular docking, Lipinski, receptor, ADME - Source URL: - Dashboard article: Abstract: The advent of antibiotic-resistant bacteria has escalated the exploration of novel antidiarrheal drugs. Persicaria vivipara is a perennial herb of the Polygonaceae family, located in high-elevation regions. The present study was designed to perform phytochemical analysis and investigate the in vivo and in silico antidiarrheal potential of a methanolic extract of Persicaria vivipara roots (MEPV). Phytochemical analysis was performed using GC-MS, and the extract was characterized by total phenolic content (TPC), total flavonoid content (TFC), and anti-microbial activity analysis. For anti-diarrheal activity, Swiss albino mice were divided into five groups. Group II was treated with loperamide, and groups III–V received 100, 200, and 400 mg per kg MEPV in castor oil, respectively, and gastrointestinal motility was assessed. Furthermore, Swiss ADME and molecular docking with the mu-opioid receptor were conducted to identify potent anti-diarrheal secondary metabolites. GC-MS analysis revealed the presence of 29 secondary bioactive compounds. The methanolic extract had 124.76 ± 1.33 mg GAE per g and 34.68 ± 1.35 mg GAE per g phenolic and flavonoid content, respectively. The antimicrobial activity tests showed 11 mm and 12 mm inhibition zones against S. aureus and E. coli as compared to ciprofloxacin (28 mm). The methanolic extract substantially reduced the distance covered by charcoal markers and inhibited diarrhea (p < 0.0001 and p < 0.001). In castor oil-induced diarrheal tests, MEPV reduced the onset of diarrhea, fecal weight, and fecal number compared with the control group (p < 0.000 to p < 0.01). SwissADME was used to identify phytocompounds that follow the Lipinski rule of five, which were subjected to docking studies. 5,5,11,11-Tetramethyltricyclo\[6.2.1.0,1,6\]undec-6-en-2-one and methyl 3-(acetyloxymethyl)biphenylene-2-carboxylate showed the highest binding affinities against opioid receptors 4DKL and 6DDE with scores of −9.6 and −7.8 kcal mol−1, respectively. Consequently, this study gives insight into the use of Persicaria vivipara extract for the treatment of diarrhea. ## Vortioxetine enhances the anti-pancreatic cancer efficacy of gemcitabine through reducing MAOB expression - Source: Frontiers in Pharmacology (journals) - Date: 2026-09-08T00:00:00Z - Categories: Docking & Screening - Journal: Frontiers in Pharmacology - DOI: 10.3389/fphar.2026.1856439 - External ID: e2370d1ad261b250ea2f6d82d39b2c5bd54ff659 - Keywords: molecular docking - Source URL: - Dashboard article: Abstract: Pancreatic cancer is a highly aggressive malignancy with a poor prognosis, largely owing to chemoresistance and frequent recurrence following gemcitabine-based therapy. This study investigated whether vortioxetine, a multimodal serotonin (5-HT) modulator, could enhance the antitumor efficacy of gemcitabine and explored the underlying molecular mechanisms. The effects of vortioxetine alone and in combination with gemcitabine were evaluated in pancreatic cancer cells and xenograft models. Cell proliferation, apoptosis, migration, invasion, and epithelial-mesenchymal transition (EMT) were assessed using functional assays and relevant molecular markers. RNA sequencing was performed to identify potential molecular targets of vortioxetine, followed by molecular docking, thermal stability assays, and gene knockdown experiments for mechanistic validation. Vortioxetine significantly enhanced the inhibitory effects of gemcitabine on pancreatic cancer cell proliferation in vitro and tumor growth in xenograft models. The combination treatment markedly promoted apoptosis, as indicated by increased levels of cleaved poly(ADP-ribose) polymerase (PARP) and cleaved caspase-3. It also suppressed cancer cell migration and invasion and reversed EMT-related changes, as demonstrated by the upregulation of E-cadherin and downregulation of N-cadherin and Snail. RNA sequencing identified monoamine oxidase B (MAOB; EC 1.4.3.4) as a potential molecular target associated with vortioxetine treatment, which was further supported by molecular docking and thermal stability assays. Moreover, MAOB knockdown phenocopied the effects of vortioxetine by enhancing gemcitabine sensitivity and suppressing EMT-associated phenotypes. Vortioxetine potentiates the antitumor efficacy of gemcitabine against pancreatic cancer, with its effects associated with reduced MAOB expression, enhanced apoptosis, and suppression of EMT. These findings support further investigation of vortioxetine combined with gemcitabine as a potential therapeutic strategy for overcoming chemoresistance in pancreatic cancer. ## ∆-Machine Learning for LC-DFT-level Excitation Energies of Bacteriochlorophyll Molecules in a LH2 Complex - Source: ChemRxiv (preprints) - Date: 2026-09-08T00:00:00Z - Authors: Sayan Maity, Vivin Vinod, Peter Zaspel, Ulrich Kleinekathöfer - DOI: 10.26434/chemrxiv.15002714/v2 - External ID: 10.26434/chemrxiv.15002714/v2 - Keywords: DFT, density functional theory, molecular descriptors - Source URL: - Dashboard article: Abstract: Bacteriochlorophyll (BChl) molecules are the key pigments in the light-harvesting complex 2 (LH2) of purple bacteria, driving solar energy conversion. During this process, fluctuations in excitation energies of the pigments at ultrafast timescales, modulated by the surrounding protein environment, govern the excitation dynamics within the complex. Because of the relatively large size of these molecules with 85 atoms and their complex electronic structure, these energy fluctuations have so far been estimated using semi-empirical methods or low-level time-dependent density functional theory (TD-DFT). In this work, we introduce a ∆-machine learning protocol to predict excitation energies at the computationally demanding long-range-corrected (LC) TD-DFT level within a quantum mechanics/molecular mechanics (QM/MM) framework, using a semi-empirical tight-binding analogue (TD-LC-DFTB) as a low-cost reference. Excitation energies for 7.2 million BChl geometries from trajectories of the LH2 complex of Rs. molischianum have been determined at the TD-LC-DFT level, both with and without taking the protein environment into account. Unexpectedly, environmental descriptors did not improve performance, so the final model relies solely on molecular descriptors. The resulting LC-DFT and machine-learned excitation energy fluctuations yield reorganization energies closer to experiment than those based on LC-DFTB. The transferability across BChl binding pockets within the protein is, however, limited. ## This Week In Cheminformatics: Issue \#037 - Source: This Week in Cheminformatics (Substack) (feeds) - Date: 2026-09-07T22:40:46+00:00 - Categories: Blog - Keywords: Cheminformatics - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Introducing Searching Synthon Spaces by Shape - Source: RDKit blog (Greg Landrum) (feeds) - Date: 2026-09-07T22:00:00+00:00 - Categories: Blog - Source URL: - Dashboard article: - Abstract: not stored for this record. ## The Accuracy Paradox: Empirical Diagnostic of Default Decision Thresholds in Multi-Label Enzyme Commission Prediction \[With Code\] - Source: arXiv (preprints) - Date: 2026-09-07T19:00:17Z - Categories: Property Prediction - Authors: Bilal Ahmad, Rajed Mehmood - External ID: 2609.07897v1 - Keywords: Enzyme - Source URL: - Dashboard article: - PDF: Abstract: Automated prediction of Enzyme Commission (EC) numbers plays a central role in functional annotation and computational drug discovery. However, standard multi-label machine learning pipelines frequently rely on default decision thresholds (t=0.50), assuming balanced prior distributions across target heads. In this study, we present a systematic empirical diagnostic of uncalibrated fixed decision boundaries operating under severe class imbalance across N = 14,096 annotated compounds categorized into six primary EC classes (EC1-EC6). Our results highlight a pronounced Accuracy Paradox: while the multi-label system achieves a deceivingly high mean accuracy of 77.16%, the macro F1-score (0.3976) and macro recall (0.3872) reveal severe predictive breakdown. Majority target classes suffer from hyper-sensitivity and over-prediction, whereas minority classes exhibit sharp recall decay, culminating in a total decision boundary collapse for EC6 (Recall = 0.00%) despite underlying discriminative power (ROC-AUC = 0.5857). Feature correlation analysis further reveals high linear redundancy among topological indices relative to fingerprint density metrics. Ultimately, this diagnostic study demonstrates that standard point predictions mask critical errors in bioinformatics workflows. We establish target-specific threshold optimization and post-hoc conformal calibration as essential, open-source post-processing safeguards for reliable applied machine learning and deep learning architectures. ## Chem(o)info radar - Source: Side Reactions (feeds) - Date: 2026-09-07T15:34:01+00:00 - Categories: Blog - Source URL: - Dashboard article: - Abstract: not stored for this record. ## dock-postprocess - Source: Macs in Chemistry (Macinchem Blog) (feeds) - Date: 2026-09-07T14:46:58+00:00 - Categories: Blog - Source URL: - Dashboard article: - Abstract: not stored for this record. ## A Dual-Key Lock Strategy for Bright Fluorescent Proteins Fabrication via Synergistic Noncovalent Activation of Aggregation-Induced Emission Luminogen - Source: Analytical Chemistry (journals) - Date: 2026-09-07T00:00:00Z - Categories: Docking & Screening - Journal: Analytical Chemistry - DOI: 10.1021/acs.analchem.6c04324 - External ID: cc83bcd58515e2558f8d4c78fa4f157fd4d2e894 - Keywords: molecular docking, bioactivity, binding affinity - Source URL: - Dashboard article: Abstract: The development of simple, bio-friendly strategies to engineer bright fluorescent proteins (FPs) is crucial for biosensing and bioimaging. Conventional synthesis of FPs requires time-consuming chromophore maturation and tedious preparation, while covalent conjugation methods often involve long reactions and risk loss of bioactivity. Herein, we propose a noncovalent “Dual-Key Lock” strategy and demonstrated the strategy using a selective aggregation-induced emission luminogen (AIEgen), TCBPE. This mechanism relies on the synergistic action of hydrogen bonding and hydrophobic interactions, which collectively confine TCBPE within the binding pocket of bovine serum albumin (BSA@TCBPE), effectively restricting intramolecular motion to activate AIE. Controlled studies with two reference AIEgens (TPE, TCPE) and molecular docking simulations validated this synergistic action with TCBPE. Based on strong noncovalent binding affinity (Kd = 54.8 nM), the fluorescence intensity of BSA@TCBPE reached 92% of its maximum value at 5 min and exhibited 12.62-fold enhancement over free TCBPE. BSA@TCBPE had a high quantum yield of 73.74% and excellent photostability under irradiation at 2 W/cm2 for 90 min. Importantly, the noncovalent conjugation preserved the native functionality of BSA, enabling its effective use in biosensing, while also facilitating high-contrast cellular imaging with low cytotoxicity. This work elucidated key supramolecular interactions and established a simple and efficient platform for constructing high-performance fluorescent proteins with biofunctional applications. ## A retrieval-augmented agent bridges virtual screening and automated synthesis - Source: ChemRxiv (preprints) - Date: 2026-09-07T00:00:00Z - Categories: Design de novo, Reaction Informatics, LLMs & Agents - Authors: Shan Zhu, Xin Chen, Pinqi Wang, Quan Jiang, Mengzhu Li, Qimeng Wang, Guoyun Zhao, Yuan Qi, Fenglei Cao, Li-Cheng Xu - DOI: 10.26434/chemrxiv.15008409/v1 - External ID: 10.26434/chemrxiv.15008409/v1 - Keywords: virtual screening, retrosynthetic prediction, LLM, generative model, synthesis planning, retrosynthesis, retrosynthetic, reaction conditions - Source URL: - Dashboard article: Abstract: Translating virtually designed drug molecules into executable synthesis protocols is essential for connecting computational discovery with automated synthesis. However, most retrosynthesis methods focus on precursor generation, offer limited route diversity, and lack the detailed reaction conditions required for executable protocols. Here, we present Syn-RRAG, a retrieval-augmented LLM-driven synthesis agent connecting virtual screening with automated synthesis and validation. Syn-RRAG integrates initial retrosynthetic prediction, reaction-precedent retrieval and hierarchical refinement to generate structured protocols with complete experimental parameters. Its local generative model searches over complete reaction hypotheses rather than relying solely on token-level beam search, enabling exploration of feasible routes with distinct reaction classes and bond disconnections. It achieved strong retrosynthesis performance on two patent-derived reaction datasets. Syn-RRAG further generated executable plans for three virtually screened drug candidates, all of which were synthesized and validated on an automated platform. These results establish a workflow connecting virtual molecular design, synthesis planning and experimental validation. ## Activity Landscape Roughness Anticipates Machine-Learning Reliability Across Environmental Chemistry Endpoints - Source: Environmental Science & Technology (journals) - Date: 2026-09-07T00:00:00Z - Categories: Property Prediction - Journal: Environmental Science & Technology - DOI: 10.1021/acs.est.6c06820 - External ID: cf108e65793ceae4d4bb80f538660b499befae1d - Keywords: QSAR, chemicals - Source URL: - Dashboard article: Abstract: Machine learning increasingly supports chemical risk assessment under REACH, TSCA, and OECD QSAR frameworks, but random-split performance can be optimistic for structurally novel chemicals. We ask whether task difficulty can be anticipated before model training. EnvMolBench spans 45 datasets, 114,500+ end point-specific records (55,984 unique structures), and 6500+ model–dataset combinations from 13 algorithms and 6 representation categories. Training-set activity landscape roughness, quantified by nearest-neighbor disagreement rate (classification) or the Structure–Activity Landscape Index (regression), was associated with best-observed performance (Spearman ρ = −0.71, 95% CI \[−0.94, −0.26\], p = 0.0009 for classification; ρ = −0.47, 95% CI \[−0.81, −0.04\], p = 0.019 for regression), surviving family-wide FDR correction for classification (q = 0.0085) but not regression (q = 0.063). On high-roughness end points, excluding activity-cliff compounds post hoc raised AUC by a mean 0.134 and lowered standardized RMSE by 0.067, whereas removing cliffs from training did not help. The three evaluated pretrained models did not outperform well-tuned baselines on the ten roughest end points. Roughness therefore provides an empirical, representation-relative diagnostic of achievable performance rather than a fundamental ceiling, motivating a workflow linking it to method selection, applicability-domain assessment, and conformal calibration. EnvMolBench, its datasets, splits, and baselines are released openly. ## Adaptive Data Splitting Strategy to Enhance Knowledge Learning for Phosphorescence Lifetime Prediction - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-07T00:00:00+00:00 - Categories: Cheminformatics, LLMs & Agents - Authors: Jingxuan Shi, Shangming Li, Yuhang Tang, Jiachi Yang, Yanzhi Guo, Songran Yang, Xuemei Pu - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c02086 - Keywords: XGBoost, LLMs, molecular features - Source URL: - Dashboard article: Abstract: Room-temperature phosphorescence (RTP) lifetime is a critical property for optoelectronic and bioimaging applications, yet its experimental measurement is labor-intensive and datasets remain scarce. Herein, we construct RTPLD, a curated dataset of 681 RTP molecules collected from extensive literature with the assistance of large language models (LLMs), providing a standardized foundation for data-driven modeling of RTP systems. To enable reliable learning from this limited but chemically diverse dataset, we propose DMFsplit, an adaptive data splitting strategy that integrates the global molecular features and local structural variations. By dynamically balancing the two-level features, DMFsplit generates more representative training sets and improves chemical knowledge acquisition in machine learning models. Extensive benchmarks across ten machine learning models demonstrate that DMFsplit consistently outperforms random and three local structure-based data splitting strategies. Coupling with the complementary ECMA feature representation, DMFsplit-XGBoost achieves R2 values of 0.84 and 0.75 on the independent and external test sets, respectively. In addition, SHAP analysis further reveals key structural motifs governing the RTP lifetime, guiding rational experimental design. Collectively, this work establishes a publicly available dataset, a generalizable data splitting strategy, and a reliable predictive model, facilitating knowledge-driven RTP research. ## Adaptive Data Splitting Strategy to Enhance Knowledge Learning for Phosphorescence Lifetime Prediction - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-07T00:00:00Z - Categories: LLMs & Agents - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c02086 - External ID: 8955979385aeef3a007a3c6ebb1a848429a770ce - Keywords: XGBoost, LLMs, molecular features - Source URL: - Dashboard article: Abstract: Room-temperature phosphorescence (RTP) lifetime is a critical property for optoelectronic and bioimaging applications, yet its experimental measurement is labor-intensive and datasets remain scarce. Herein, we construct RTPLD, a curated dataset of 681 RTP molecules collected from extensive literature with the assistance of large language models (LLMs), providing a standardized foundation for data-driven modeling of RTP systems. To enable reliable learning from this limited but chemically diverse dataset, we propose DMFsplit, an adaptive data splitting strategy that integrates the global molecular features and local structural variations. By dynamically balancing the two-level features, DMFsplit generates more representative training sets and improves chemical knowledge acquisition in machine learning models. Extensive benchmarks across ten machine learning models demonstrate that DMFsplit consistently outperforms random and three local structure-based data splitting strategies. Coupling with the complementary ECMA feature representation, DMFsplit-XGBoost achieves R2 values of 0.84 and 0.75 on the independent and external test sets, respectively. In addition, SHAP analysis further reveals key structural motifs governing the RTP lifetime, guiding rational experimental design. Collectively, this work establishes a publicly available dataset, a generalizable data splitting strategy, and a reliable predictive model, facilitating knowledge-driven RTP research. ## Adaptive Discovery of Uncharted Zeolitic Imidazolate Frameworks - Source: ChemRxiv (preprints) - Date: 2026-09-07T00:00:00Z - Authors: Yen-hsu Lin, Zichao Rong, Mei-Yan Gao, Sara Alnasser, Omar M. Yaghi - DOI: 10.26434/chemrxiv.15008404/v1 - External ID: 10.26434/chemrxiv.15008404/v1 - Source URL: - Dashboard article: Abstract: Discovering new zeolitic imidazolate framework (ZIF) structures remains a major challenge because their assembly depends on the nature of the linker and synthesis conditions. Here, we introduce linker width as a design parameter for ZIF synthesis. We designed 4-substituted benzimidazole linkers with expanded width and combined them with smaller linkers in an adaptive machine-learning (ML) high-throughput strategy. This adaptive discovery scheme yielded nine new mixed-linker ZIFs, including six adopting the previously unrealized bcv topology. Single-crystal structural analysis reveals ordered linker distributions in which the wider 4-substituted benzimidazolate linkers code for the formation of 10-membered-ring needed for the apertures of the bcv cage. Comparison with representative mixed-linker ZIFs further identifies the width of linkers and the extension ratio as steric descriptors of topology selection. These results establish chemically interpretable design principles for mixed-linker ZIF assembly and demonstrate how linker design coupled with ML-guided synthesis can access unexplored regions of chemical space. ## An Explainable Machine Learning-Based QSAR Framework for Predicting Thrombin Inhibitory Activity - Source: Pharmaceuticals (journals) - Date: 2026-09-07T00:00:00Z - Categories: Cheminformatics, Property Prediction - Journal: Pharmaceuticals - DOI: 10.3390/ph19091411 - External ID: 36e97e4a4b901165e3288e1427ae217b00feaf1f - Keywords: Morgan fingerprint, QSAR, bioactivity, IC50 - Source URL: - Dashboard article: Abstract: Background/Objectives: Public thrombin bioactivity records contain heterogeneous endpoints, replicate measurements, related chemical series, and potentially reactive compounds that may bias quantitative structure–activity relationship models. In this study, we developed an explainable, assay-aware, and leakage-safe machine learning framework for predicting thrombin-inhibitory activity. Methods: Exact Ki and IC50 records for human thrombin (CHEMBL204) were standardized, converted to pActivity, and aggregated using predefined criteria. The final dataset comprised 5189 unique compounds represented by development-filtered Mordred descriptors and Morgan fingerprint. The models were optimized using development-only out-of-fold validation and evaluated using random and scaffold-disjoint held-out tests. Results: In the random-split analysis, ConsensusAll achieved an out-of-fold R2 of 0.7588 and a held-out test R2 of 0.7480, with an RMSE of 0.7360, MAE of 0.5307, and concordance correlation coefficient of 0.8540. In the scaffold-disjoint locked test, ConsensusTop5 achieved R2 = 0.5831, RMSE = 0.9484, and MAE = 0.7290, respectively. The applicability domain covered 92.68% of the locked test compounds and yielded R2 = 0.6041. One hundred Y-randomization runs produced a mean R2 of −0.1233 (empirical p = 0.0099). Conclusions: The framework provides useful predictions within the represented chemical space and measurable generalization for unseen scaffolds. This supports compound prioritization, although prospective biochemical validation remains necessary. ## Analysis of Senolytics Sophoretin and Dasatinib across Alzheimer's disease and Annulus Fibrosus Degeneration - Source: Journal of Integrative and Translational Biomedicine (journals) - Date: 2026-09-07T00:00:00Z - Categories: Docking & Screening - Journal: Journal of Integrative and Translational Biomedicine - DOI: 10.67316/jintb.v1.i3.53 - External ID: 025f0e3e642c5bd95440df84869982e90431bfc2 - Keywords: molecular docking, AutoDock Vina - Source URL: - Dashboard article: Abstract: Background: Flavonoid sophoretin (quercetin) exhibits multi-target biological activity, including senolytic-adjunct that complements the effects of senolytic dasatinib. This study evaluates sophoretin inhibitory and molecular binding with cholinesterase, network interactions, and effects in degenerative biological systems. Methods: In vitro inhibition of cholinesterase was assessed by Ellman's method, and in silico molecular docking was performed using AutoDock Vina. Putative molecular targets of sophoretin were predicted using SwissTargetPrediction and intersected with Alzheimer's-associated proteins to identify overlapping targets, which were mapped for protein–protein network interaction and subjected to enrichment analysis. Effects of sophoretin were evaluated using annulus fibrosus tissue treated with a dasatinib–sophoretin cocktail, as a degeneration-sensitive peripheral model. Results: Sophoretin showed concentration-dependent inhibition of cholinesterase, with lower potency than donepezil at low-to-intermediate concentrations but comparable inhibition. Binding affinities were nearly equivalent to donepezil, with sophoretin interaction profile distinguished by a greater contribution of hydrogen bonds/polar contacts, contrasting with donepezil predominantly hydrophobic binding mode. Network analysis identified 25 overlapping targets between sophoretin and Alzheimer's-associated proteins, with significant enrichment in MAPK and PI3K-Akt signaling, amyloid response, and apoptosis regulation. Gene expression analysis of annulus fibrosus tissue following combination treatment revealed 170 differentially expressed genes, indicating broad and bidirectional transcriptional reprogramming. Conclusion: Sophoretin exhibits multi-scale biological activity with Alzheimer's-relevant protein networks and systemic transcriptional responsiveness in a degeneration-sensitive tissue model. These findings support the characterization of sophoretin as a multi-modal bioactive compound with potential relevance across Alzheimer's-associated and broader degenerative pathways, warranting further in vivo validation. ## Anti-Inflammatory Activity of Monoterpene Phenol Dimers from Satureja bachtiarica - Source: ACS Omega (journals) - Date: 2026-09-07T00:00:00+00:00 - Categories: Docking & Screening - Authors: Simone Di Micco, Milena Masullo, Noemi Marigliano, Ester Colarusso, Marzieh Rahmani Samani, Anna Schettino, Anella Saviano, Francesco Maione, Sonia Piacente, Giuseppe Bifulco - Journal: ACS Omega - DOI: 10.1021/acsomega.6c03194 - Keywords: molecular docking, molecular dynamics - Source URL: - Dashboard article: Abstract: Satureja bachtiarica (Bakhtiari savory) is an aromatic herb cultivated both as a culinary spice and for its medicinal uses. Species of Satureja are known to exhibit anti-inflammatory properties. Monoterpene phenol dimers with biphenyl skeletons have been identified in these plants and structural investigations revealed the key role of the biphenyl scaffold in interacting with MAPEG family proteins. Thus, in the present contribution, we investigated by biochemical and cell-based assays, together with molecular docking and molecular dynamics investigations, the modulation of inflammatory mediators of three isolated monoterpene phenol dimers (1–3). Molecular modeling and in vitro enzymatic assays of 1–3 against soluble epoxide hydrolase (sEH) and 5-lipoxygenase (5-LOX) identified the structural elements accountable for binding vs the biological targets. Unlike 2 and 3, in vitro experiments revealed that 1 did not show cytotoxicity. Interestingly, 1 selectively alters the profile of AA-derived mediators produced by LPS-activated macrophages, suppressing both IL-6 and PGE2. The obtained results revealed how compound 1 may not only suppress inflammation but also promote the active resolution of inflammatory responses and, by integrating computational analysis, provide clues to design more effective analogues. ## AquaTox-Predictor: An MMoE-Enhanced Multimodal Deep Learning Framework for Aquatic Ecotoxicity Prediction - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-07T00:00:00+00:00 - Categories: Cheminformatics, Property Prediction, ADMET & Safety - Authors: Jingwei Zhang, Le Xiong, Fei Pan, Weihua Li, Xinxin Yu, Wenxiang Song, Dian Sheng, Yun Tang, Guixia Liu - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c02770 - Keywords: molecular fingerprints, classification model, chemicals - Source URL: - Dashboard article: - Code: Abstract: As the concentration of compounds in aquatic environments rises, the degradation of aquatic habitats underscores the increasing importance of studying the impact of chemicals on different aquatic populations. In this study, we developed AquaTox-Predictor, a multitask, multimodal classification model that integrates molecular fingerprints, molecular graphs, and a Multi-gate Mixture-of-Experts (MMoE) framework for prediction of seven aquatic ecotoxicities of chemicals. The results of the validation sets demonstrated that our model outperformed other methods, achieving the highest average AUC with an improvement of up to 5.75% compared to existing approaches. Our method can correctly identify the substructures that define molecular properties as well as the differences and relationships between end points, thus enhancing interpretability. These results demonstrate that AquaTox-Predictor can achieve high accuracy in predicting various aquatic toxicities and conduct interpretability analysis on compounds, which is conducive to its application in research on other more complex toxicity end points. All the code and datasets are freely available online at https://github.com/jwZhang-lab/AquaTox-Predictor. ## AquaTox-Predictor: An MMoE-Enhanced Multimodal Deep Learning Framework for Aquatic Ecotoxicity Prediction - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-07T00:00:00Z - Categories: Cheminformatics - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c02770 - External ID: 38547a777a0567371539c05a7f22683a986dfced - Keywords: molecular fingerprints, classification model, chemicals - Source URL: - Dashboard article: - Code: Abstract: As the concentration of compounds in aquatic environments rises, the degradation of aquatic habitats underscores the increasing importance of studying the impact of chemicals on different aquatic populations. In this study, we developed AquaTox-Predictor, a multitask, multimodal classification model that integrates molecular fingerprints, molecular graphs, and a Multi-gate Mixture-of-Experts (MMoE) framework for prediction of seven aquatic ecotoxicities of chemicals. The results of the validation sets demonstrated that our model outperformed other methods, achieving the highest average AUC with an improvement of up to 5.75% compared to existing approaches. Our method can correctly identify the substructures that define molecular properties as well as the differences and relationships between end points, thus enhancing interpretability. These results demonstrate that AquaTox-Predictor can achieve high accuracy in predicting various aquatic toxicities and conduct interpretability analysis on compounds, which is conducive to its application in research on other more complex toxicity end points. All the code and datasets are freely available online at https://github.com/jwZhang-lab/AquaTox-Predictor. ## Artificial Intelligence in Scalable Materials Synthesis and Manufacturing - Source: AI Chemistry (journals) - Date: 2026-09-07T00:00:00Z - Journal: AI Chemistry - DOI: 10.3390/aichem1030014 - External ID: eed61c39045d806ab82e63d15de6136d02293814 - Keywords: reinforcement learning, property prediction, graph neural networks - Source URL: - Dashboard article: Abstract: While Artificial Intelligence (AI) has transformed materials discovery, the primary bottleneck to technological impact remains the transition from lab-scale synthesis to robust, industrial-scale manufacturing. Most promising materials perish in this depth, which is referred as the “valley of death”. The current review consolidates and critically evaluates the emerging ecosystem of AI-driven strategies and frameworks designed specifically to bridge this “lab-to-fab” gap. The review shifts our attention from property prediction to the engineering-driven problems of manufacturability. Furthermore, the review discusses the main obstacles to scaling the production of materials, such as reproducibility, process optimization in the context of uncertainty, and techno-economic viability, as well as the AI approaches being developed to overcome them. This includes Natural Language Processing (NLP) for method extraction, graph neural networks for reaction modeling, reinforcement learning for process control, Bayesian optimization for definition of process windows, and integrated AI–Techno-Economic Analysis (TEA) frameworks. Equally importantly, the review examines the principal failure modes that constrain practical deployment, including out-of-distribution generalization, incomplete and non-transferable literature-derived data, simulator-to-plant mismatch, uncertainty miscalibration, and the continued need for expert oversight. The article concludes with a forward-looking roadmap for the future of AI in chemical and materials engineering. The proposed conclusion is defined by a paradigm shift from simply finding new materials to creating viable, economical, and scalable pathways to produce them, thereby enabling a new era of synthesis-aware materials innovation. ## C-MAGE: Automated Extraction and Organization Engine for Molecular Representations in Chemical Documents - Source: ChemRxiv (preprints) - Date: 2026-09-07T00:00:00Z - Categories: Cheminformatics - Authors: Alexander Taylor, Jeehyun Hwang, Mihir Surve, Wei Wang, Olaf Wiest - DOI: 10.26434/chemrxiv.15008460/v1 - External ID: 10.26434/chemrxiv.15008460/v1 - Keywords: SMILES, chemical space coverage, Molecular Representations - Source URL: - Dashboard article: Abstract: To exploit the data present in chemical literature for AI applications in chemistry, an expanded and reliable extraction pipeline is needed. AI-ready chemical databases broaden chemical space coverage and provide benchmarking and training sets for machine learning model development to drive innovation within the field. To curate data from chemical documents a multimodal approach is needed. C-MAGE is an open-source molecule extraction engine that processes figures and tables in PDF files to provide machine-readable molecule datasets of the structures as CXS-MILES strings, an encompassing representation which evades heuristic translation of superatoms to SMILES subunits. This led to a 7.1% increase in accurate molecular representations generated. All data extracted undergo confidence-based sorting to ensure high precision (77.6%) in dataset collection, and removes the need for manually sorting datapoint validity. Both the CXSMILES and their confidence values are reproducible and are generated efficiently (1.7 seconds per molecule). C-MAGE’s use is demonstrated by the generation of an annotated dataset containing 1454 CXSMILES strings from 38 chemical documents. C-MAGE thus enables large-scale information extraction of chemical structures in a consistent and reproducible pipeline, complementing the more common text extractions. ## CampChem: Rubric-Grounded Adaptive Campaigns for Multi-Round Organic Synthesis with Residual Spectral Observation - Source: ChemRxiv (preprints) - Date: 2026-09-07T00:00:00Z - Categories: Cheminformatics, Reaction Informatics, LLMs & Agents - Authors: Zixuan Shi, Mengyao Qian, Yichao Sun, Haotian Gu, Jiarui Feng - DOI: 10.26434/chemrxiv.15008453/v1 - External ID: 10.26434/chemrxiv.15008453/v1 - Keywords: SMILES, LLM, Mistral, reinforcement learning, retrosynthesis - Source URL: - Dashboard article: Abstract: Organic synthesis is a multi-round campaign, not a one-shot SMILES completion: a failed solvent, a poisoned catalyst, or leftover starting-material peaks in the 1H NMR force the chemist to revise conditions and try again. Tool-using chemistry agents and condition classifiers still emit a single recipe and then stop. We propose CampChem, an adaptive campaign agent built on a chemistry-instruction LLM. A longitudinal campaign adaptive planner (LCAP) keeps a short memory of failed condition slots and peak signatures and decodes the next catalyst/solvent/reagent tuple under a mask. Chemistry rubric-as-judge reinforcement learning (Chem-RaJ) scores stoichiometry, hazard compatibility, slot completeness, and spectral consistency instead of using a free-form LLM judge. A residual spectral booster (RSB) injects a permutation-invariant peak-set residual when NMR is available and gates to zero otherwise. On SMolInstruct, CampChem reaches 67.1% forward-synthesis and 36.1% retrosynthesis Top-1 exact match with a Mistral-7B LoRA backbone. On USPTO-Condition it attains 0.2940 overall Top-1, and on AutoLabs Experiment 5 it raises protocol F1 from 0.75 to 0.81. Ablations show that Chem-RaJ carries most of the spectrum-free gain, LCAP matters once a prior tuple has been rejected, and RSB contributes mainly on simulated NMR revision. ## Computationally Guided Discovery of Cannabinoid Derivatives for Antiviral and Anticancer Applications: From Chemical Space to Experimental Validation - Source: ChemRxiv (preprints) - Date: 2026-09-07T00:00:00Z - Categories: Property Prediction, Docking & Screening, ADMET & Safety, Design de novo - Authors: VIVEK KUMAR YADAV - DOI: 10.26434/chemrxiv.15008399/v1 - External ID: 10.26434/chemrxiv.15008399/v1 - Keywords: virtual screening, molecular docking, drug likeness, QSAR, quantum chemistry, pharmacokinetic, ADMET, receptor, molecular dynamics, MD simulations, chemical diversity - Source URL: - Dashboard article: Abstract: Cannabinoids represent a chemically diverse and structurally tunable class of natural and synthetic compounds that have attracted growing interest as molecular scaffolds for therapeutic discovery. However, their extensive chemical diversity, polypharmacology, conformational flexibility, and complex pharmacokinetic properties make the systematic identification and optimisation of therapeutically useful derivatives challenging. Recent advances in computer-aided drug discovery (CADD), molecular simulation, quantum chemistry, quantitative structure–activity relationship (QSAR) modelling, and artificial intelligence (AI) provide opportunities to interrogate the chemical space of cannabinoids in a systematic and mechanistically informed manner. This review examines computational strategies for the discovery and optimisation of cannabinoid derivatives with potential antiviral and anticancer applications, emphasising the integration of complementary computational approaches rather than reliance on isolated in silico predictions. A hierarchical computational workflow is described, progressing from cannabinoid chemical-space generation and ligand preparation to disease-relevant target identification, virtual screening, molecular docking, molecular dynamics (MD) simulations, binding free-energy estimation, quantum-chemical analysis, and ADMET/drug-likeness prediction. For antiviral discovery, computational investigations of SARS-CoV-2 targets, including the main protease (Mpro/3CLpro), papain-like protease (PLpro), spike protein, and spike–ACE2 interface, are discussed alongside emerging host-directed targets. For anticancer applications, computational studies involving EGFR, HER2, VEGFR, PI3K/AKT, MAPK, apoptosis, autophagy, oxidative stress, and cell-cycle regulation are evaluated. Particular emphasis is placed on the complementary roles of docking and MD simulations in characterising binding modes, interaction persistence, conformational stability, and molecular recognition, while MM-PBSA/MM-GBSA and related approaches provide additional energetic criteria for candidate prioritisation. QSAR and AI/ML-assisted approaches are further considered as complementary tools for expanding chemical-space exploration, predicting biological activity and ADMET properties, and enabling multi-parameter candidate prioritisation. Importantly, computational affinity scores are not treated as sufficient evidence of target inhibition or therapeutic efficacy. Instead, the review emphasises a progressive validation framework in which computationally prioritised candidates are evaluated through biochemical target-engagement studies, cellular assays, and, where appropriate, in vivo models. Experimental observations can subsequently inform structure–activity relationship analysis, lead optimisation, and iterative computational rescreening, establishing a closed-loop strategy for integrating molecular recognition, dynamic behaviour, binding energetics, pharmacokinetic properties, toxicity, and biological activity during cannabinoid lead development. The review critically examines key limitations of current computational approaches, including incomplete treatment of ligand flexibility, scoring-function limitations, protein conformational heterogeneity, limited experimental validation, dataset bias in AI/ML models, and translational challenges associated with cannabinoid solubility, bioavailability, metabolism, receptor promiscuity, psychoactive liabilities, and off-target effects. Collectively, current evidence indicates that computationally guided cannabinoid discovery is a promising but predominantly preclinical strategy. Further progress will depend on integrating CADD, molecular simulation, AI-assisted molecular design, uncertainty-aware prediction, and experimental validation into iterative multi-objective optimisation frameworks that jointly address potency, selectivity, pharmacokinetics, toxicity, and translational relevance. ## Computer-aided drug design of quinoxaline based selective oestrogen receptor α modulators for breast cancer therapy: molecular docking, MD simulation, MM-PBSA and ADME/T analysis. - Source: SAR and QSAR in environmental research (journals) - Date: 2026-09-07T00:00:00Z - Categories: Docking & Screening, ADMET & Safety, Free Energy & MD - Journal: SAR and QSAR in environmental research - DOI: 10.1080/1062936X.2026.2719503 - External ID: 05606e4613fa2ce1eac6f36f5dd58b3870fa986c - Keywords: molecular docking, virtual screening, PubChem, ADME, ADMET, pharmacokinetic, binding free energy, receptor, MD simulation, MD simulations - Source URL: - Dashboard article: Abstract: Oestrogen receptor alpha (ERα) is a driver of hormone-dependent breast cancer, yet current therapies are often hindered by drug resistance and adverse effects. In this study, we developed a structure-based virtual screening workflow to identify novel quinoxaline derivatives as computationally prioritized potential ERα modulators. The quinoxaline analogues obtained from PubChem are screened through validated docking model of ERα. The top-ranked candidates were further refined by MD simulations in GROMACS for 300 ns to assess complex stability and interaction persistence, followed by MM-PBSA binding free energy calculations to quantify binding energetics. The prioritized quinoxaline hits exhibited more favourable predicted docking scores than reference ligand, tamoxifen, while MD trajectory analyses indicated stable complex formation with sustained predicted interactions involving key ERα binding site residues. MM-PBSA highlights LIG3 as the most promising lead with a favourable energy (ΔG = -34.15 kcal/mol). Complementary ADMET profiling predicted favourable pharmacokinetic behaviour and low toxicity for the prioritized compounds. This integrated in silico strategy demonstrates that quinoxaline scaffolds, specifically LIG3, represent promising computationally prioritized templates for the further development and experimental evaluation of ERα targeted ligands. These findings provide a reliable computational framework for the prioritization and structural optimization of next-generation anti-breast cancer agents. ## Convergence Mechanisms of Generative Models in Molecular Conformational Sampling - Source: Journal of Chemical Theory and Computation (journals) - Date: 2026-09-07T00:00:00+00:00 - Authors: Nagesh B E, Jagannath Mondal - Journal: Journal of Chemical Theory and Computation - DOI: 10.1021/acs.jctc.6c01009 - Keywords: Transformer, Generative Models, denoising, Diffusion models - Source URL: - Dashboard article: Abstract: Characterizing equilibrium conformational ensembles with deep generative models requires understanding whether a model reproduces a target distribution and how it reaches that distribution. Here, we compare two generative routes to molecular conformational sampling, stochastic relaxation and deterministic transport, using denoising diffusion probabilistic models and rectified-flow models across systems of increasing complexity: a multimodal two-dimensional potential, the folded miniprotein Trp-cage, and a high-dimensional dihedral representation of an intrinsically disordered protein. We show that these paradigms differ in end point fidelity and in how distributional error is resolved during sampling. Diffusion models converge through pronounced late-stage stochastic relaxation and robustly recover the configurational breadth across neural architectures. Rectified flow approaches the target distribution through deterministic transport and therefore depends more strongly on architectural expressivity, particularly in heterogeneous, high-dimensional landscapes. Entropy and moment-evolution analyses further show that diffusion more reliably restores the ensemble location and fluctuation structure, whereas rectified flow requires Transformer-level feature mixing to represent transport geometry accurately. These results establish the convergence mechanism as a practical design principle for molecular generative sampling, clarifying when stochastic diffusion provides robustness and when deterministic transport requires higher representational capacity. ## Critical benchmarking of machine-learned interatomic potentials for intermolecular and noncovalent interactions - Source: Machine Learning: Science and Technology (journals) - Date: 2026-09-07T00:00:00+00:00 - Authors: Kamal Nayal, IlKwon Cho, Olexandr Isayev - Journal: Machine Learning: Science and Technology - DOI: 10.1088/2632-2153/aea39f - Keywords: MACE, MLIP, DFT - Source URL: - Dashboard article: Abstract: Accurate benchmarking of intermolecular interaction energies is central to evaluating quantum chemical methods and guiding the development of reliable machine-learned interatomic potentials (MLIPs). We benchmark five MLIPs (AIMNet2(2023), AIMNet2(2025), MACE-OFF23(M), MACE-OMol, and UMA-S-OMol) across twenty-one datasets spanning hydrogen-bonded, dispersion- and pi-dominated, sigma-hole, ionic and charge transfer, and repulsive nonequilibrium interactions, with reference values at or near CCSD(T)/CBS accuracy. AIMNet2(2025) is a continually pretrained variant of AIMNet2(2023) that retains the original architecture but incorporates 3.8 million additional structures curated to improve noncovalent interactions (NCIs). AIMNet2(2025) improves on its predecessor across nearly all benchmark categories, with the largest gains in the hydrogen-bonded, sigma-hole, and repulsive regimes, while remaining competitive with the much larger MACE-OMol and UMA-S-OMol. The supramolecular S12L and L7 benchmarks show only marginal improvement: every evaluated MLIP exhibits large errors driven by a small number of pathological complexes. Two factors beyond intrinsic model quality significantly influence the reported performance. First, partial overlap between training and benchmark data, quantified here via systematic overlap detection, inflates apparent accuracy for all models, most strongly for those trained on OMol25. Second, differences in the DFT reference level used for MLIP training establish irreducible error floors, so superior benchmark performance may partly reflect closer proximity of the training functional to the CCSD(T)/CBS reference rather than stronger modeling capability. Sigma-hole interactions emerge as the category with the lowest training-benchmark overlap across all models and therefore provide the most discriminating test of true generalization. Meaningful MLIP evaluation must account for data provenance, reference theory consistency, and the distinction between interpolation and out-of-distribution generalization, particularly as standard NCI benchmark sets become absorbed into large-scale training datasets. ## De Novo Design of Targeted Drugs by Precisely Programming Multiple Molecular Binding - Source: Journal of the American Chemical Society (journals) - Date: 2026-09-07T00:00:00+00:00 - Authors: Chuangyuan Zhao, Yu Jin, Xue Li, Zhengming Shi, Wenhao Shi, Sai Ge, Dongsheng Liu, Yuanchen Dong - Journal: Journal of the American Chemical Society - DOI: 10.1021/jacs.6c15828 - Keywords: De Novo Design, binding affinity - Source URL: - Dashboard article: Abstract: Targeted drug discovery has enabled the development of effective therapies for many diseases. However, a fundamental challenge remains in translating molecular binding into functional performance for the development of novel targeted drugs. Here, we present a programmable DNA nanoframe (NF) strategy that converts ligand binding into robust biological functions through precise spatial organization of multiple ligands. By encoding target dimensions and binding geometry, we have successfully constructed potential targeted nanodrugs against diverse proteins, which exhibit high avidity with markedly enhanced binding affinity and selectivity. A design-driven blockade of VEGF has also been translated into marked antiangiogenic efficacy both in vitro and in vivo. Overall, the programmable DNA NFs establish a modular and generalizable paradigm for targeted drug design, upgrading weak or non-functional binders into effective inhibitors through rational spatial programming. ## De Novo Design of Targeted Drugs by Precisely Programming Multiple Molecular Binding - Source: Journal of the American Chemical Society (journals) - Date: 2026-09-07T00:00:00Z - Journal: Journal of the American Chemical Society - DOI: 10.1021/jacs.6c15828 - External ID: 36eaf923c0ebec1eeb9dbbf67c74e3c010b0e341 - Keywords: De Novo Design, binding affinity - Source URL: - Dashboard article: Abstract: Targeted drug discovery has enabled the development of effective therapies for many diseases. However, a fundamental challenge remains in translating molecular binding into functional performance for the development of novel targeted drugs. Here, we present a programmable DNA nanoframe (NF) strategy that converts ligand binding into robust biological functions through precise spatial organization of multiple ligands. By encoding target dimensions and binding geometry, we have successfully constructed potential targeted nanodrugs against diverse proteins, which exhibit high avidity with markedly enhanced binding affinity and selectivity. A design-driven blockade of VEGF has also been translated into marked antiangiogenic efficacy both in vitro and in vivo. Overall, the programmable DNA NFs establish a modular and generalizable paradigm for targeted drug design, upgrading weak or non-functional binders into effective inhibitors through rational spatial programming. ## De Novo Linear Peptides as Potential Recognition Elements and/or Aggregation Modulators of Amyloid-β42 in Alzheimer’s Disease - Source: ChemRxiv (preprints) - Date: 2026-09-07T00:00:00Z - Authors: C. Rocha, G. Martins, A. Vieira, G. Simplício, S. Varanda, F. Fernandes, N. Galamba - DOI: 10.26434/chemrxiv.15008448/v1 - External ID: 10.26434/chemrxiv.15008448/v1 - Keywords: force fields, molecular dynamics - Source URL: - Dashboard article: Abstract: The development of deep learning algorithms for de novo protein design has opened the way for novel in silico strategies to develop peptide binders targeting protein aggregation implicated in neurodegenerative diseases. When combined with molecular simulations, these approaches provide a promising framework for investigating protein-peptide recognition and its impact on aggregation. Herein, we studied, through molecular dynamics, four de novo peptides of sizes ranging between 7 and 15 residues, generated with RFdiffusion, combined with ProteinMPNN and AlphaFold, using different constrained sequences of the Aβ42 peptide and a common hotspot region (residues 16-22). The shorter peptides formed cross-β sheet structures with the hotspot sequence, whereas the longer peptides adopted α-helical conformations. The hotspot sequence in trans and the segment Aβ42(16-20) linked to an hexalysine sequence were also investigated. Aβ42 was modeled using CHARMM36m (C36m) and AMBER99SB-DISP force fields, with the latter showing an aggregation propensity comparable to C36m under 8 M urea denaturing conditions. The peptides’ potential to modulate aggregation was assessed for C36m Aβ42 model tetramers at 1:1 peptide-to-monomer ratios. Although none of the peptides prevented the formation of all possible oligomeric species (i.e., dimer, trimer, and tetramer), several, selectively inhibited or delayed tetramer formation within the simulation timescale, resulting in less compact assemblies. The underlying mechanisms differed among peptides, with some altering intermolecular β-sheet organization through persistent interactions with the C-terminal region and others influencing tetramer formation with limited participation in βsheet interactions. The behavior of the peptides 16-22 (KLVFFAE) and the 16-20-KKKKKK closely reproduced experimental observations. At high temperatures none of the peptides precluded tetramerization, suggesting that enhanced hydrophobic interactions outweigh peptide-mediated inhibition. Notably, charged peptides showed the strongest effects on tetramer formation, indicating an important role for electrostatic interactions in inhibiting Aβ42 aggregation. ## Decoding cell–cell communication in spatial transcriptomics: mechanistic insights, modeling constraints, and analytical caveats - Source: Briefings in Bioinformatics (journals) - Date: 2026-09-07T00:00:00+00:00 - Authors: Yuesong Wu, Haohao Su, Yuehua Cui - Journal: Briefings in Bioinformatics - DOI: 10.1093/bib/bbag477 - Keywords: receptor - Source URL: - Dashboard article: Abstract: Cell–cell communication (CCC) is essential for maintaining tissue organization and driving biological progression, yet its inference from transcriptomic data has long been limited by the absence of spatial context. Advances in spatial transcriptomics (ST) now enable mechanistically grounded analyses of CCC by preserving the physical organization of cells and their microenvironments. In this review, we examine recent methodological developments in CCC inference from ST data, focusing on how statistical, optimal transport, and deep learning frameworks incorporate spatial information to model ligand–receptor (LR) interactions and downstream signaling. We also summarize key mechanism-driven components shared across spatial and non-spatial CCC approaches. In addition, we discuss how tissue heterogeneity and spatial architecture can introduce context-dependent biases, particularly for permutation-based inference, and outline mechanistic considerations such as LR biochemistry, signal transduction, and condition-specific communication. We further highlight databases that curate intercellular conduction and intracellular signaling processes. By integrating spatial constraints with biochemical and computational principles, this review offers an integrated assessment of the opportunities and limitations of current approaches. We conclude by identifying key methodological challenges and future directions for developing robust, scalable, and mechanistically interpretable CCC inference as ST technologies continue to advance. ## Deep3DCCS: Geometry-Driven Collision Cross Section Prediction from Multi-View Molecular Projection Tensors - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-07T00:00:00+00:00 - Categories: Cheminformatics, Spectra & Analytical - Authors: Pattipong Wisanpitayakorn, Vivek Bhakta Mathema, Kwanjeera Wanichthanarak, Thanyanatthamon Keawhawong, Narumol Jariyasopit, Pairash Saiviroonporn, Trongtum Tongdee, Yongyut Sirivatanauksorn, Sakda Khoomrung - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c00371 - Keywords: SMILES, convolutional neural - Source URL: - Dashboard article: Abstract: Collision cross section (CCS) measurements from ion mobility–mass spectrometry improve confidence in molecular identification in metabolomics and lipidomics and reduce false-positive annotations in untargeted analyses. However, experimental CCS coverage remains incomplete, and accurate computational prediction remains needed. Here, we present Deep3DCCS, a geometry-driven deep learning framework that predicts CCS from stacked multi-angle molecular projection tensors using a three-dimensional convolutional neural network operating across the projection-view and two spatial image dimensions. Three-dimensional molecular structures generated from SMILES strings were rotated around the x, y, and z-axes, converted into binary two-dimensional projections, and assembled into standardized multi-view tensors for CCS regression. Deep3DCCS was developed and evaluated using curated AllCCS and METLIN lipid CCS datasets containing 3948 molecule–adduct entries across \[M – H\]−, \[M + H\]+, and \[M + Na\]+ ions, and input optimization identified 32 × 32 pixel projections with five rotations per axis as a practical balance between prediction accuracy, model stability, and computational feasibility. In independent true-validation experiments, AllCCS-trained models achieved MRPE values of 3.27–3.79% on AllCCS validation data and 2.93–3.72% on METLIN validation data, whereas METLIN-trained models transferred less effectively to the broader AllCCS chemical space. Overall, these results support multi-view molecular projection tensors as a viable geometry-driven representation for CCS prediction and demonstrate that substantial CCS-relevant information can be learned directly from three-dimensional molecular morphology using a 3DCNN architecture, with low single-digit prediction errors across independent and cross-dataset validation experiments. ## Design and Structure of Protein Cages Based on Helical Fusion and Machine Learning - Source: Journal of the American Chemical Society (journals) - Date: 2026-09-07T00:00:00+00:00 - Authors: Pablo San Segundo-Acosta, Johanne Le Coq, Jasminka Boskovic, Robin A. Aglietti, Peter Bowers, Todd O. Yeates, Roger Castells-Graells - Journal: Journal of the American Chemical Society - DOI: 10.1021/jacs.6c11491 - Source URL: - Dashboard article: Abstract: Self-assembling protein cages are versatile nanoscale architectures with broad applications in drug delivery, vaccine development, and structural biology. Historically, two main strategies have been used to construct such cages: genetic fusion of oligomeric domains connected by helical linkers, and computational interface design using either physics-based or machine learning-based methods. Here, we extend the original fusion approach using modern AI algorithms and more sophisticated treatments of helix bending to create protein cages with novel architectures composed exclusively of trimeric building blocks arranged in tetrahedral symmetry. Of 15 designs tested experimentally, multiple sequence variants of two of these designs assembled predominantly into soluble, monodisperse particles of the expected size, with native molecular masses of 633 kDa (T33-Fus-1A, B) and 638 kDa (T33-Fus-2). Cryo-electron microscopy (cryo-EM) structures of three distinct sequence variants spanning from 3.0 to 3.9 Å in resolution confirmed the intended structures in atomic detail, with C-α RMSD values over the entire assemblies as low as 2 Å. The predicted modes of helix bending were similarly validated. The results highlight the impact of methodological improvements for achieving a level of regularity and design precision that has largely evaded prior applications of the fusion approach. These findings expand the prospects and accessible design space for self-assembling protein nanomaterials. ## Design and syntheses of 4-aminoquinoline-thiazolidinone hybrids as anti-protozoal agents - Source: Frontiers in Chemistry (journals) - Date: 2026-09-07T00:00:00Z - Categories: Docking & Screening, ADMET & Safety - Journal: Frontiers in Chemistry - DOI: 10.3389/fchem.2026.1892807 - External ID: 094fec0d5bbcaa00ec1d0bf488e7ec9060f4c48d - Keywords: molecular docking, ADMET prediction, enzyme, molecular dynamics, MD simulation, ADMET - Source URL: - Dashboard article: Abstract: Molecular hybridization is an effective approach employed to synthesize molecules that possess the ability to function as dual inhibitors, thereby addressing the issue of growing resistance. Resistance by the malaria parasite is one such example. Therefore, the study aims to develop hybrid compounds consisting of 4-aminoquinoline and 4-thiazolidinone, utilizing molecular docking for sorting, followed by synthesis and evaluation of their anti-malarial potential. The molecules were selected on the basis of affinity from the computational study. To strengthen the in silico docking, a 200ns molecular dynamics simulation study was also performed. The title compounds, 7-18, were synthesized in a three-step reaction, involving non-polar solvent addition reaction resulting in heterocyclization. The synthesis was followed by characterization of all the synthesized compounds, and they were further subjected to biological screening through in-vitro , and the selected actives to in-vivo evaluation. An acute toxicity study was also conducted. The compounds exhibited promising action, with five compounds, 8, 9, 14, 15, and 18, demonstrating efficacy comparable to the reference drug in the Pf -DHFR enzyme inhibition assay. Four of the active compounds selected for the Plasmodium berghei murine model displayed excellent activity, with percentage parasitaemia inhibition to the extent of 84%, and mean survival time up to 27 days. The MD simulation study showed the stability of the ligand-protein complex up to 200 ns. The in silico ADMET prediction, along with acute toxicity testing of the most active compound, showed no major toxicity. The study concludes the hybrid compounds to be promising antimalarial agents, and the scaffold may be a good lead for the development of next-generation antimalarials. ## Design, synthesis, molecular docking and antiparasitic evaluation of a novel metronidazole derivative against clinical isolates of Entamoeba histolytica - Source: Journal of Parasitic Diseases (journals) - Date: 2026-09-07T00:00:00Z - Categories: Docking & Screening - Journal: Journal of Parasitic Diseases - DOI: 10.1007/s12639-026-01998-1 - External ID: b09e2ceb42bd6c5663021aa4f0e0ce7e91fe42c7 - Keywords: molecular docking - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Development of interpretable machine learning models for predicting the probability of sepsis in patients with pulmonary fibrosis in the intensive care unit: based on MIMIC-IV and multi-database validation - Source: Frontiers in Cellular and Infection Microbiology (journals) - Date: 2026-09-07T00:00:00Z - Journal: Frontiers in Cellular and Infection Microbiology - DOI: 10.3389/fcimb.2026.1894489 - External ID: 0342205b97b91216dea70c29920af8deca558455 - Keywords: Gradient Boosting, XGBoost - Source URL: - Dashboard article: Abstract: Limited by small sample size, single-institution design, and insufficient comprehensive external validation across heterogeneous healthcare systems, no study to date has systematically validated the predictive performance of machine learning models for sepsis occurrence in an intensive care unit (ICU) population with concomitant pulmonary fibrosis through multiple large-scale databases. This retrospective multi-database study utilized two large databases to establish and validate a machine learning model for predicting the probability of sepsis occurrence in ICU patients with pulmonary fibrosis. In this study, 542 patients from the MIMIC-IV database were divided into a training set (381 patients) and an internal validation set (161 patients) in a 7:3 ratio, and external validation was performed on the MIMIC-III (186 patients) database. Six machine learning algorithms were employed: Decision Tree (DT), Extreme Gradient Boosting (XGBoost), Logistic Regression (LR), Lightweight Gradient Boosting Machine (LightGBM), Support Vector Machine (SVM), and Artificial Neural Network (ANN). Baseline variables were screened using least absolute shrinkage and selection operator (Lasso) regression to identify potential predictors. The interpretability of the model was evaluated using Shapley Additive Explanations (SHAP) analysis. The entire cohort consisted of 728 ICU patients with pulmonary fibrosis. We identified nine consistently crucial clinical characteristics, including gender, dementia, pneumonia, antibiotics, nephrotoxic drugs, glucocorticoids, sequential organ failure assessment (Sofa) score, red blood cell distribution width, and total serum calcium. The ANN algorithm performed optimally, with an area under the curve (AUC) of 0.878 in the training set, 0.837 in the internal validation set, and 0.857 in the MIMIC-III external validation set. SHAP analysis indicated that Sofa was the most influential predictor, followed by antibiotics and pneumonia. Additionally, a web tool was developed to facilitate the prediction of sepsis probability in clinical practice. This study is the first to develop and validate a machine learning model for predicting sepsis in ICU patients with pulmonary fibrosis across multiple databases. The ANN model, combined with SHAP interpretability, provides a reliable decision-making tool for clinical decision support, and its consistency has been verified in two databases, including our internal validation cohort. ## DIME: Dynamics Inferred from Monte Carlo Ensembles via Continuous-Time Markov Chains - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-07T00:00:00+00:00 - Categories: Targets & Structures, Free Energy & MD - Authors: Krishna Praneet Mulukutla, Marimuthu Krishnan - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c01396 - Source URL: - Dashboard article: Abstract: Monte Carlo (MC) sampling and related enhanced sampling methods can efficiently explore the conformational space of complex biomolecular systems and accurately estimate equilibrium populations of conformational states. However, the absence of temporal information prevents these methods from directly describing molecular kinetics. Here we present a framework for constructing a continuous-time Markov chain (CTMC) from a static, Boltzmann-weighted ensemble, enabling the generation of physically plausible kinetic trajectories without requiring extensive molecular dynamics (MD) simulation. The method includes (i) clustering the MC ensemble into metastable discrete states whose populations define the stationary distribution, (ii) building a rate matrix over these states that satisfies detailed balance by design, with transition rates set by the free energy barriers separating the states, estimated directly from the free energy surface of the ensemble, and (iii) calibrating the absolute time scale by matching the slowest relaxation mode to a short MD reference trajectory. We validate the approach across six molecular systems spanning a broad range of complexity: n-pentane and alanine dipeptide as low-dimensional benchmarks; a WLALL pentapeptide; the achiral peptide AIB9, whose left- and right-handed helices are related by an exact symmetry that provides a reference-free test of the inferred kinetics; wild-type chignolin as a β-hairpin folder; and the millisecond native-state dynamics of bovine pancreatic trypsin inhibitor (BPTI) from a long reference trajectory. In each case the inferred model reproduces MD-derived rate matrices, mean first-passage times, and implied time scales following a short calibration step. Since the method relies on the free energy surface, it can equally well be applied to surfaces obtained from enhanced sampling simulations, as demonstrated for alanine dipeptide using well-tempered metadynamics. The method provides a principled route to kinetic insight when MD convergence is prohibitively expensive but thermodynamic sampling is tractable and serves as a complement to existing Markov state model approaches that build kinetic models from MD data. ## Electrophile-Free, Self-Immolative Linker-Enabled Near-Infrared Ratiometric Fluorescent Probe for High-Fidelity, Nonperturbative Imaging of Acetylcholinesterase (AChE) Dynamics in Epileptic Mice - Source: Analytical Chemistry (journals) - Date: 2026-09-07T00:00:00Z - Categories: Docking & Screening - Journal: Analytical Chemistry - DOI: 10.1021/acs.analchem.6c04296 - External ID: 04020638c69206a7c66c1470895085e48301fdbb - Keywords: Molecular docking, binding affinity, enzyme - Source URL: - Dashboard article: Abstract: Fluorescence monitoring of acetylcholinesterase (AChE) levels is critical for evaluating neurotransmitter regulation; however, nonspecific background hydrolysis, poor targeting fidelity, and single-channel readout severely compromise quantitative accuracy. Conventional electrophilic leaving groups (quinone methides and aldehydes) deplete intracellular nucleophiles (cysteine or glutathione), thereby inducing oxidative stress and disrupting redox homeostasis. Herein, a series of near-infrared (NIR) ratiometric fluorescent probes (Cy667, Cy701, and Cy777) have been rationally designed that integrate a polarity-reversing nucleophilic self-immolative linker, a specific AChE recognition unit, and an ammonium-assisted AChE-targeting moiety. The well-regulated self-immolative nucleophilic cleavage completely eliminates undesired side reactions and electrophilic byproduct interference while maintaining intracellular redox homeostasis without exhausting glutathione (GSH). The optimal probe, Cy701, exhibits exceptional selectivity (>100-fold over other hydrolases) and sensitivity (LOD = 4.9 mU/mL) toward AChE, enabling a ratiometric signal log(F636/F832) in the NIR region. Molecular docking calculations verify that Cy701 has an excellent binding affinity for AChE, which enables real-time visualization of endogenous AChE in PC12 and HepG2 cells. Notably, Cy701 displays superior blood–brain barrier (BBB) permeability and favorable intracellular retention. It has been robustly implemented for in vivo NIR ratiometric imaging of AChE dynamics in acute and chronic epileptic mouse models, unambiguously discriminating pathological states from healthy conditions with a signal-to-background ratio (SBR) of 10.0 (P < 0.001). This work constitutes a universal design paradigm for high-fidelity, nonperturbative enzyme-activatable fluorescent probes and offers a robust scaffold for accurate AChE monitoring and in-depth neurobiological dissection of epilepsy. ## Empagliflozin Targets NF-κB Signaling Through PTGS2 and TLR4 in Polycystic Ovary Syndrome: A Drug Repurposing Study and Molecular Simulation - Source: Journal of Xenobiotics (journals) - Date: 2026-09-07T00:00:00Z - Categories: Docking & Screening - Journal: Journal of Xenobiotics - DOI: 10.3390/jox16050169 - External ID: e8d6dab47525469ebd98a64db9cafed06c7d0037 - Keywords: molecular docking, binding affinity, molecular dynamics - Source URL: - Dashboard article: Abstract: Polycystic ovary syndrome (PCOS) is a multifactorial endocrine disorder characterized by chronic inflammation, insulin resistance, and reproductive dysfunction. Although empagliflozin, a sodium-glucose cotransporter-2 inhibitor, has demonstrated anti-inflammatory and metabolic benefits, its molecular mechanisms in PCOS remain poorly understood. This study investigated the anti-inflammatory mechanisms of empagliflozin in PCOS using network pharmacology, molecular docking, and molecular dynamics simulations. Potential drug targets were identified using SwissTargetPrediction and SuperPred, while PCOS- and inflammation-related genes were obtained from GeneCards. Overlapping targets were subjected to Gene Ontology, Kyoto Encyclopedia of Genes and Genomes, protein–protein interaction network, and hub gene analyses. Molecular docking and 50 ns molecular dynamics simulations were performed to evaluate binding affinity and complex stability. Key inflammatory targets identified included TNF, IL6, IL1B, TLR4, STAT3, and PTGS2, with significant enrichment in cytokine-mediated signaling, TNF signaling, and NF-κB pathways. Empagliflozin showed strong binding affinities for PTGS2 (−9.0 kcal/mol) and TLR4 (−8.8 kcal/mol), while molecular dynamics simulations demonstrated stable protein–ligand complexes throughout the simulation. These findings suggest that empagliflozin may alleviate PCOS-associated inflammation by modulating the TLR4/NF-κB/PTGS2 signaling axis, supporting its potential as a repurposed therapeutic agent for PCOS and providing a foundation for future experimental validation. ## Ensemble Machine Learning for Thickness‐Dependent Bandgap Prediction in Perovskite Solar Cell Thin Films - Source: physica status solidi (a) (journals) - Date: 2026-09-07T00:00:00Z - Categories: Cheminformatics - Journal: physica status solidi (a) - DOI: 10.1002/pssa.70518 - External ID: cc7e094b0b9722dd1e9975d97e94bea0b44bcfc3 - Keywords: RDKit, Random Forest, XGBoost, molecular descriptors - Source URL: - Dashboard article: Abstract: This report presents a comprehensive ensemble machine learning study for predicting the Bandgap × Thickness product (eV·nm), a key opto‐electronic figure of merit for perovskite solar cell absorber layers. A dataset of device records is featurized using physics‐informed descriptors encoding device architecture‐related composition fractions with layer‐stack interactions, RDKit molecular descriptors for perovskite organic A‐site cations (methylammonium, formamidinium) with halide anions as fraction‐weighted averages, and Mordred extended descriptor vectors for the same chemical components. Six ensemble regressors are trained and benchmarked as Random Forest, Extra Trees, XGBoost, LightGBM, a Voting Ensemble, and a Stacking Ensemble with a Ridge meta‐learner. Among them, XGBoost achieves the highest test R 2 of 0.98 (RMSE = 14.28 eV·nm and MAE = 5.98 eV·nm), while the Stacking Ensemble delivers the lowest MAE of 4.79 eV·nm. Feature attribution via SHAP, MDI, and permutation importance (PI) consistently identifies perovskite thickness, the engineered effective absorbance term (BG × Thickness × PCE), and the Br/I halide mixing ratio as the dominant predictors. Mordred descriptors contributed ~20.5% of the total MDI importance to demonstrate molecular fingerprinting at device‐level datasets. ## Epigallocatechin Gallate (EGCG) Inhibits African Swine Fever Virus Replication by Targeting Key Viral Proteins and Regulating Host Lipid Metabolism. - Source: Emerging microbes & infections (journals) - Date: 2026-09-07T00:00:00Z - Categories: Docking & Screening - Journal: Emerging microbes & infections - DOI: 10.1080/22221751.2026.2731508 - External ID: 2117b97ded8418edf29874fae2a2a83481598dda - Keywords: Molecular docking, kinase - Source URL: - Dashboard article: Abstract: AbstractAfrican swine fever (ASF) is a highly pathogenic swine infectious disease caused by African swine fever virus (ASFV), with a mortality rate approaching 100% in domestic pigs and causing severe economic losses to the global pig industry. Despite the recent approval of live-attenuated ASF vaccines in limited regions such as Vietnam, universally safe, globally authorized commercial vaccines and effective antiviral therapeutics remain unavailable, creating an urgent demand for innovative anti-ASFV intervention strategies. In this study, epigallocatechin gallate (EGCG) was identified to exert prominent inhibitory effects on ASFV proliferation in vitro. Viral life cycle assays indicated that EGCG mainly targets the internalization and post-entry genome replication stages. Molecular docking combined with bio-layer interferometry (BLI) confirmed high-affinity direct binding between EGCG and two indispensable ASFV proteins, p72 and p1192R. Furthermore, EGCG dose-dependently activates the AMP-activated protein kinase (AMPK) signaling pathway, downregulates lipid synthesis-related genes, and reverses ASFV-induced lipid droplet accumulation and abnormal increases in total cholesterol (TC), triglycerides (TG), and free fatty acids (FFAs), thereby disrupting the lipid metabolic microenvironment required for viral replication. Collectively, EGCG suppresses ASFV replication through dual mechanisms: targeting key viral proteins and regulating host lipid metabolism. This study highlights EGCG as a promising anti-ASFV candidate and offers a novel theoretical basis for the development of anti-ASFV agents. ## Essential oils and plant extracts for sustainable crop protection - Source: Academia Biology (journals) - Date: 2026-09-07T00:00:00Z - Journal: Academia Biology - DOI: 10.20935/acadbiol8506 - External ID: 0cf1a4a46f03985e7782cc6643a5ac4e73ae0ac0 - Keywords: Molecular docking - Source URL: - Dashboard article: Abstract: In recent years, environmental issues related to the extensive use of synthetic pesticides have attracted great attention to finding sustainable methods of plant protection. Plant-derived materials, particularly essential oils (EOs) and plant extracts, are considered to be potential sources of natural compounds because of their rich chemical composition and wide range of biological activities. This review summarizes scientific evidence published between January 2015 and May 2026 on the chemical composition and biological activities of plant-derived products against phytopathogenic bacteria, fungi, insects, and weeds. Across the studies reviewed, plant-derived products showed antibacterial, antifungal, insecticidal, phytotoxic, and antioxidant effects. Essential oils showed biological activity in many of the included studies, but their activity cannot be directly compared with that of nonvolatile extracts because the available studies differ considerably in the organisms tested, experimental methods, formulations, concentrations, exposure times, and units used to report activity. Biological activities have been associated with several mechanisms, including membrane disruption, oxidative stress generation, metabolic disturbances, and interference with neuronal signalling pathways. Activity against plant pathogens has also been reported in some in vivo, greenhouse, postharvest, and field studies, although evidence from field trials is still scarce. Molecular docking studies have proposed possible interactions between individual EO constituents and molecular targets involved in bacterial virulence or insecticidal activity, but these predictions still require experimental confirmation. Despite these encouraging findings, practical use of plant-derived biopesticides is still limited by chemical variability, lack of standardized methodologies, insufficient field trials, and formulation challenges. Improved formulations, including nanotechnology-based and controlled-release systems, may help overcome some of these limitations and facilitate their use in sustainable agriculture. ## Exploring the Substrate Selectivity of the Condensation Domain VibH through Engineered Adenylation Domains, Non-Native Amide Synthesis, and Substrate Recognition Analysis. - Source: ACS chemical biology (journals) - Date: 2026-09-07T00:00:00Z - Categories: Docking & Screening - Journal: ACS chemical biology - DOI: 10.1021/acschembio.6c00537 - External ID: dc178287cbdea4cd0ee4012c7de9306590e0581c - Keywords: Molecular docking - Source URL: - Dashboard article: Abstract: Nonribosomal peptide synthetases (NRPSs) are modular assembly-line enzymes that generate structurally diverse and biologically active natural products. Although adenylation (A) domain selectivity has been extensively characterized and engineered, the specificity of condensation (C) domains, which catalyze peptide bond formation and can constrain NRPS reprogramming, remains less well understood. Here, we report a systematic functional analysis of VibH, a stand-alone C domain VibH from the vibriobactin biosynthetic pathway. By integrating VibH with wild-type and engineered variants of the upstream aryl acid A domain EntE, we bypassed intrinsic A-domain constraints and independently interrogated donor- and acceptor-site substrate tolerance. The donor site of VibH exhibited stringent specificity, accepting only mono- and disubstituted benzoic acid derivatives closely resembling 2,3-dihydroxybenzoic acid (DHB), whereas bulkier aryl substrates were not processed. In contrast, the acceptor site displayed broad tolerance toward structurally diverse amines. Monoamines showed clear chain-length dependence, with productive turnover restricted to medium-chain substrates, whereas diamines were more broadly accepted. Systematic analysis further revealed distance-dependent steric tolerance, in which bulky substituents were accommodated when positioned distal to the reactive amine but were restricted beyond an upper steric threshold. This engineered reconstitution strategy enabled the synthesis of numerous non-native amide conjugates and uncovered an asymmetric substrate-recognition architecture characterized by a stringent donor site and a permissive yet spatially constrained acceptor site. Molecular docking analysis provided a structural rationale for these trends by suggesting distinct binding modes for native and non-native donor substrates and distance-dependent accommodation of bulky amine acceptors. These findings provide mechanistic insights into C-domain specificity and establish a framework for rational NRPS reprogramming. ## Framework-Defined Microenvironments Differentiate Molecular CO2 Reduction Sites - Source: ChemRxiv (preprints) - Date: 2026-09-07T00:00:00Z - Categories: Free Energy & MD - Authors: Chenxi Li, Mårten S. G. Ahlquist - DOI: 10.26434/chemrxiv.15008446/v1 - External ID: 10.26434/chemrxiv.15008446/v1 - Keywords: free energy perturbation, DFT - Source URL: - Dashboard article: Abstract: Molecular catalysts offer well-defined active sites for electrochemical CO2 reduction to CO, while immobilization in metal–organic frameworks (MOFs) can increase catalyst loading and organize these sites within a porous environment. This strategy often assumes that repeated molecular sites remain functionally equivalent after framework incorporation. Here, using DFT-benchmarked, site-resolved empirical valence bond–free energy perturbation (EVB–FEP) simulations in an explicit electrode-supported Co-porphyrin MOF, we studied four elementary CO2-to-CO reactions across eight matched boundary and inner sites and four progressively assembled electrochemical environments. Framework embedding produces the dominant energetic redistribution and establishes a persistent boundary–inner differentiation in CO2 adsorption. We identify its microscopic origin as a site-dependent local axial electric field generated by different balances of solvent, ion, and neighboring-framework electrostatics. Boundary and inner sites therefore stabilize the same reaction through different microscopic environments. Regional CO2 mobility further shows that intrinsic reactivity and transport accessibility provide separate constraints on site function. Together, these results show that framework incorporation creates a spatial distribution of functionally distinct catalytic microenvironments rather than simply multiplying an equivalent molecular catalyst. By combining high densities of well-defined molecular active sites with spatially heterogeneous yet synthetically tunable environments, framework materials provide a bridge between molecular and heterogeneous catalyst behavior. Once the microscopic origins of site-dependent reactivity are captured by suitable descriptors, this mechanistic understanding can inform materials screening and guide framework design toward balancing reaction energetics and accessibility. ## From Chemical Scaffold to Predicted Toxicity: A Scaffold-Aware Pipeline for Reliable In Silico Prioritization of Heracleum Furanocoumarins - Source: International Journal of Molecular Sciences (journals) - Date: 2026-09-07T00:00:00Z - Categories: Cheminformatics, Docking & Screening - Journal: International Journal of Molecular Sciences - DOI: 10.3390/ijms27177947 - External ID: 3f39e2d6848d27f244d928b0dbefb90812090aa4 - Keywords: molecular docking, PubChem - Source URL: - Dashboard article: Abstract: In silico methods generate numerous toxicity predictions for natural compounds, yet their mutual consistency and transferability to new structures are rarely verified, and the predictions are often taken at face value. Using furanocoumarins of the genus Heracleum as a model class of natural products, we developed a reproducible computational pipeline that evaluates not the predictions themselves but their reliability. Two datasets were analyzed: an extended reference set (DS1, 2008 compounds from PubChem) and a taxonomically verified natural set (DS2, 102 Heracleum compounds). The pipeline combined scaffold analysis, molecular docking against 44 off-target proteins, redocking, prediction of ten phenotypic toxicity endpoints, scaffold-resolved separability testing, and scaffold-aware machine learning. The chemical space was concentrated around two scaffolds, and the descriptor space was non-randomly organized with respect to them along all 13 axes (false discovery rate (FDR)-adjusted p < 0.01), indicating that the predicted toxicological profile is determined predominantly by molecular structure. The mean docking score corresponded to only the first principal component of the multidimensional binding profile, and the ten endpoints were largely independent of one another. Critically, scaffold separability did not predict transferability: structurally determined signals were only partially reproduced on unseen scaffolds, and to differing degrees across endpoints. A multi-criteria prioritization identified 21 top-priority candidates for experimental follow-up. The reliability of in silico assessments for this class cannot be assumed without scaffold-aware validation. ## Identification of key genes and metabolites in thyroid eye disease through integrated transcriptomic and metabolomic analysis - Source: Frontiers in Endocrinology (journals) - Date: 2026-09-07T00:00:00Z - Categories: Docking & Screening - Journal: Frontiers in Endocrinology - DOI: 10.3389/fendo.2026.1844082 - External ID: 1ce4b54f31672c7b3e2c0013ba999a5c2f50bd53 - Keywords: molecular docking - Source URL: - Dashboard article: Abstract: Thyroid eye disease (TED) markedly compromises ocular function and quality of life, imposing significant healthcare and economic burdens. Present therapeutic strategies, which mainly include glucocorticoids, teprotumumab, immunosuppressive drugs, and surgical treatments, exhibit limited effectiveness and are associated with high rates of relapse. Consequently, unraveling the molecular mechanisms of TED and discovering new diagnostic and therapeutic targets hold immense clinical importance. In this study, we performed an in-depth analysis of the transcriptomic and metabolomic landscapes of orbital adipose tissue in individuals with TED, aiming to fill the gap in comprehensive molecular characterizations and metabolic perturbations in this disease. By adopting a multi-omics integration strategy, we leveraged transcriptome sequencing, broad-spectrum metabolomics, weighted gene co-expression network analysis (WGCNA), machine learning, immune cell infiltration profiling, molecular docking, QPCR, and immunohistochemistry for biomarker identification. These techniques were systematically employed to pinpoint key genes and metabolites associated with TED, leading to the development of an accurate disease prediction model. Remarkably, our analysis of the PRJNA1314138 dataset revealed 4, 250 differentially expressed genes (DEGs), with BPIFA1 and SERPINB3 identified as crucial genes through machine learning algorithms. These genes demonstrated outstanding predictive performance in both internal \[Area under the curve (AUC)=0.953\] and external dataset validations (AUC = 0.717). Furthermore, metabolomic profiling detected 2, 158 metabolites, pinpointing three key metabolites: gluconic acid, colneleic acid, and isoleucine-methionine. The integration of transcriptomic and metabolomic data underscored the significant enrichment of the tyrosine metabolism pathway, establishing functional links between the identified genes and metabolites. These insights offer novel perspectives on the molecular foundations of TED and propose that BPIFA1 and SERPINB3 could serve as promising biomarkers, while gluconic acid, colneleic acid and isoleucine-methionine may hold potential as metabolic indicators. Our research provides a solid framework for comprehending the pathophysiology of TED. ## Identifying therapeutic target genes for acute pancreatitis based on the druggable genes. - Source: Computer methods in biomechanics and biomedical engineering (journals) - Date: 2026-09-07T00:00:00Z - Categories: Docking & Screening - Journal: Computer methods in biomechanics and biomedical engineering - DOI: 10.1080/10255842.2026.2723090 - External ID: 60c4768c6b9797340d08b2ad5a1d212fd917c3fe - Keywords: molecular docking - Source URL: - Dashboard article: Abstract: To screen therapeutic target genes for AP, MR analysis was conducted, followed by DEG screening, nomogram, GSEA, drug prediction, and molecular docking analyses.529 druggable genes were obtained. After intersection, 36 overlapping genes were identified. Eight overlapping feature genes were screened after PPI and machine learning. Among them, four therapeutic target genes were screened as therapeutic target genes, containing ACVR1C, RASGRP1, RPTOR, and SPTBN1. These four genes are mainly involved in primary immunodeficiency, and both are negatively associated with activated dendritic cells, neutrophils, and macrophages. Molecular docking analysis highlighted the affinity of the four therapeutic target genes for glucose. ## Immunomodulatory Potential of Vaginal Microbiota-Derived Metabolites in Pregnant Women: Insights from ADMET Profiling and Molecular Docking - Source: BioChem (journals) - Date: 2026-09-07T00:00:00Z - Categories: Docking & Screening, ADMET & Safety - Journal: BioChem - DOI: 10.3390/biochem6030025 - External ID: badb651461770a8b0af476cc506e787841b42dc1 - Keywords: Molecular Docking, ADMET, pharmacokinetic - Source URL: - Dashboard article: Abstract: Background: Pregnancy involves coordinated immunological adaptations that maintain maternal–fetal tolerance while preserving protection against infections. There is growing evidence that vaginal microbiota-associated metabolites contribute to immune balance, yet their molecular mechanisms remain incompletely understood. Methods: This study used an in silico approach combining absorption, distribution, metabolism, excretion, and toxicity (ADMET) profiling, molecular docking, and structural interaction analysis. Eight compounds—five organic acids, two bacteriocins, and hydrogen peroxide—were evaluated against eight immune and inflammatory protein targets relevant to pregnancy. Results: The ADMET predictions revealed favorable pharmacokinetic and safety profiles for small organic acids, particularly butyric and propionic acids. Enterocin-HF yielded the most favorable predicted binding energies across the evaluated targets, particularly TNF-α and NF-κB. Redocking of the COX-2 co-crystallized ligand reproduced the experimental binding pose with an RMSD of 1.21 Å, supporting the ability of the protocol to recover the crystallographic binding mode. Conclusions: Overall, these findings prioritize Enterocin-HF and selected organic acids for subsequent experimental evaluation. However, the predicted interactions should not be interpreted as evidence of functional immunomodulation or therapeutic efficacy. ## In silico assessment of natural compounds against mycobacterium tuberculosis Rv1509 protein - Source: Applied Biological Chemistry (journals) - Date: 2026-09-07T00:00:00Z - Categories: ADMET & Safety - Journal: Applied Biological Chemistry - DOI: 10.1186/s13765-026-01118-w - External ID: 3434909860b0bfded9c51d3ed56a7a423bdaa9a1 - Keywords: virtual screening, molecular dynamics, ADME, pharmacokinetic - Source URL: - Dashboard article: Abstract: Tuberculosis (TB) remains a leading cause of mortality globally, driven by the infectious pathogen, Mycobacterium tuberculosis (M.tb). A novel DNA methyltransferase (DNA MTase), encoded by the Rv1509 gene and involved in TB pathogenesis, has been identified as a promising therapeutic target of anti-TB drugs. The present research employs an in silico approach to identify potential inhibitors of the Rv1509-encoded DNA MTase using a computational drug design pipeline. A multi-stage virtual screening of ZINC natural compounds was conducted against Rv1509. These phytomolecules were retrieved from the ZINC database, following computationally intensive docking and analysis of Absorption, Distribution, Metabolism, and Excretion (ADME) properties, top hits with a docking score ≤ -8.0 kcal/mol and favourable predicted pharmacokinetic profiles were prioritized. Subsequently, molecular dynamics simulations (MD) and principal component analysis (PCA) were employed to corroborate these hits. Of these, ZINC00338392, ZINC01662782, ZINC04104877, and ZINC96316367 ligands exhibited hydrogen bond formation with functional residue of DNA MTase, indicating biological relevance of binding. MD analysis revealed stable protein–ligand complexes during a 200 ns simulation. These computational analyses suggest that natural compounds bind with high predicted affinity to the active site of the Rv1509-encoded DNA MTase, warranting future experimental validation of their potential as novel leads for anti-TB drug development. ## Integrated Genome Mining and Bioactivity-Guided Isolation of Antimicrobial Peptides from Bacillus amyloliquefaciens BS4. - Source: Probiotics and antimicrobial proteins (journals) - Date: 2026-09-07T00:00:00Z - Categories: Docking & Screening - Journal: Probiotics and antimicrobial proteins - DOI: 10.1007/s12602-026-11205-5 - External ID: ece3615f11a5afecffad18639674ffce201782eb - Keywords: Molecular docking, in silico screening, Bioactivity - Source URL: - Dashboard article: Abstract: Bacterial resistance remains a critical global health challenge, driving the continuous search for novel antimicrobial agents. Bacillus amyloliquefaciens is a recognized repository of bioactive metabolites; however, its full biosynthetic potential requires integrated genomic and experimental validation. This study characterized the antimicrobial profile of B. amyloliquefaciens BS4 through a hybrid pipeline. Genome sequencing and de novo assembly revealed a 3.9 Mb chromosome with a G + C content of 46.14%. Functional annotation identified 3,887 coding sequences, including pathways for siderophore biosynthesis and a complete bacilysin biosynthetic cluster. BGC analysis using antiSMASH v7.1.0 and BAGEL4 identified 18 biosynthetic gene clusters, while similarity network analysis via BiG-SCAPE highlighted unique singleton BGCs, indicating untapped biosynthetic diversity. Although in silico screening via Macrel predicted two putative cationic antimicrobial peptides (AMPs), bioactivity-guided purification utilizing sequential RP-HPLC, and de novo sequencing revealed a distinct set of four active peptides. Notably, three of these sequences were identified as fragments derived from the BclA exosporium protein family, highlighting the structural proteome as a non-canonical source of antimicrobials. The purified fractions exhibited activity against M. luteus and E. coli, while displaying no significant hemolytic activity or cytotoxicity, even above the MIC values. Molecular docking further supported the interaction of these candidates with bacterial targets. Overall, this hybrid strategy effectively uncovers the antimicrobial complexity of BS4, revealing 'cryptic' peptide candidates with therapeutic potential. ## Integration of Single-Cell and Bulk RNA Sequencing Data to Identify Lactylation-Related Gene Signatures in Hepatic Ischemia–Reperfusion Injury Using Machine Learning Algorithms - Source: International Journal of Molecular Sciences (journals) - Date: 2026-09-07T00:00:00Z - Journal: International Journal of Molecular Sciences - DOI: 10.3390/ijms27177965 - External ID: 899640b908b16e236cc2e070f47e5f141c7890ab - Keywords: ChEMBL, Random Forest - Source URL: - Dashboard article: Abstract: Hepatic ischemia–reperfusion injury (HIRI) is not only a common complication of liver transplantation and major hepatic surgery but also a critical determinant of postoperative prognosis. Lactate metabolic reprogramming has been observed in HIRI, yet the role of lactate and its related lactylation in the pathogenesis of HIRI remains unclear. To address this, we integrated single-cell and bulk RNA-seq data with multiple bioinformatic approaches. Five single-cell gene set activity scoring methods (AUCell, UCell, singscore, ssGSEA, and AddModuleScore) were applied to evaluate lactylation activity across cell types, followed by differentially expressed gene (DEG) analysis and high-dimensional Weighted Correlation Network Analysis (hdWGCNA) to identify lactylation-associated genes. Five machine learning algorithms (Random Forest, Boruta, LASSO, GBM, and Decision Tree) were used to screen optimal feature genes, with SHAP analysis further explaining their importance. Bulk RNA sequencing data from the Gene Expression Omnibus (GEO) database were used for validation. Furthermore, NR4A3-related inhibitors were screened using the ChEMBL online tool and assessed by docking and molecular dynamic simulation. We observed significant heterogeneity in lactate metabolism activity across cell types in hepatic ischemia–reperfusion injury (HIRI), with higher activity levels observed for hepatocytes and mononuclear phagocytes. The integration of SHAP and machine learning identified PFKFB3, ZYX, and NR4A3 as closely associated with high lactylation after HIRI, and cross-analysis with bulk RNA data confirmed their consistent upregulation. Candidate gene expression was experimentally validated in a murine liver IRI model through Western blotting and RT-qPCR. Although lactylation has been previously reported in HIRI, this study’s unique contribution is to reveal the cell-type heterogeneity of lactylation-related gene expression at the single-cell level through multi-omics integration and machine learning. The identification of NR4A3, PFKFB3, and ZYX as lactylation-associated regulators proposes novel therapeutic targets for improving graft survival in liver transplantation. ## Interfacial thermal transport spectroscopy: a review of theory and experiment. - Source: Physical chemistry chemical physics : PCCP (journals) - Date: 2026-09-07T00:00:00Z - Journal: Physical chemistry chemical physics : PCCP - DOI: 10.1039/d6cp02619b - External ID: 04e0d71b78fe54691b94a751b8838cfcd4a2d466 - Keywords: molecular dynamics - Source URL: - Dashboard article: Abstract: Thermal transport across interfaces is a critical bottleneck in the thermal management of modern microelectronics, particularly as devices scale toward the nanoscale with increasingly high-power densities. While bulk material properties are well understood, the physics governing heat transfer at interfaces, defined by carrier transmission and scattering, remains a complex challenge. Here, we review methods and studies on interfacial thermal transport spectroscopy, bridging fundamental theory with the state-of-the-art modelling and experiments. We first examine the fundamentals of the phonon gas model and carrier coupling, followed by a detailed discussion of computational approaches ranging from atomistic Green's functions (AGF) and molecular dynamics (MD) simulations at the nanoscale to the Boltzmann transport equation (BTE) at the microscale. Then, we evaluate popular experimental techniques, such as frequency-domain thermoreflectance (FDTR), time-domain thermoreflectance (TDTR) and electron energy loss spectroscopy (EELS), emphasizing their role in resolving spectral phonon contributions. We further highlight emerging data-driven methods and machine learning approaches that accelerate physical understanding and materials discovery. Finally, we outline open challenges in characterizing spectral interfacial thermal transport within computational and experimental frameworks, as well as the persistent gaps between them. This review aims to provide a unified perspective on understanding and optimizing interfacial heat dissipation for next-generation electronic and energy devices. ## Interpretable Machine Learning for Predicting Cytochrome P450 Mechanism-Based Inhibition - Source: ChemRxiv (preprints) - Date: 2026-09-07T00:00:00Z - Categories: Cheminformatics - Authors: Khushashish Bhasin, Sudham Engle, Devendra Kumar Dhaked - DOI: 10.26434/chemrxiv.15008396/v1 - External ID: 10.26434/chemrxiv.15008396/v1 - Keywords: MACCS keys, ECFP, RDKit, PubChem, Molecular representations, Cytochrome P450, enzyme - Source URL: - Dashboard article: Abstract: Cytochrome P450 (CYP450) enzymes play a crucial role in the metabolism of endogenous and exogenous compounds, and their inhibition can lead to clinically significant drug–drug interactions and adverse effects. Mechanism-based inhibition (MBI) is a time-dependent form of CYP450 inhibition in which enzyme-mediated bioactivation of an inhibitor can result in persistent or irreversible enzyme inactivation. Early identification of compounds with MBI liability is therefore important for reducing metabolic and drug–drug interaction risks during drug discovery. In this study, a curated dataset of 569 unique MBI compounds was compiled from the literature, while non-mechanism-based inhibitors (NMBIs) were obtained from PubChem. To construct a chemically diverse negative dataset, Butina clustering-based sampling was employed. Molecular representations based on RDKit and Mordred descriptors, MACCS keys, and ECFP fingerprints were evaluated using five machine learning algorithms, with feature selection and model optimization applied systematically. CYP-aware stratified splitting was employed to ensure appropriate representation of CYP isoform–class combinations across the training and test sets, and its performance was compared with random stratified splitting. Among the evaluated molecular representations, RDKit descriptors generally provided the strongest predictive performance, with optimized models achieving test-set ROC-AUC values of up to 0.92 in the CYP-aware evaluation. Applicability-domain analysis further demonstrated higher prediction accuracy for compounds within the defined chemical space, supporting the reliability of predictions for structurally represented compounds. In addition, structural-alert analysis identified class-associated structural motifs, including statistically significant alerts preferentially associated with MBI compounds, providing chemically interpretable information complementary to the machine learning models. Evaluation using the expanded MBI569 dataset further assessed the robustness of the predictive models across different dataset sizes and splitting strategies. Overall, this study demonstrates a systematic and interpretable approach for predicting CYP450 mechanism-based inhibition, with potential utility for the early identification of MBI liabilities during drug discovery. ## KiMecO: A Kinetic Mechanism Optimizer for Experimentally Constrained Master Equation Models - Source: The Journal of Physical Chemistry A (journals) - Date: 2026-09-07T00:00:00Z - Journal: The Journal of Physical Chemistry A - DOI: 10.1021/acs.jpca.6c04372 - External ID: d06ad831e4c60bcaa4829eb8f6cdfee8925d6625 - Source URL: - Dashboard article: Abstract: Detailed chemical kinetic mechanisms are widely used to model complex gas-phase reactions in atmospheric and combustion chemistry, but are often underconstrained. In those cases, model predictions may remain accurate due to parameter error cancellation within the validation conditions, yet fail outside of that range. We introduce KiMecO, an uncertainty informed, master equation based kinetic mechanism optimizer that constrains the rate coefficients of a submechanism against concentration–time profiles of multiple species across multiple conditions. It leverages machine learning and parallel computation to explore the sensitive master equation parameter space and yield physically consistent correlations among rate coefficients. We show that ensembles of master equation models are generally more robust than single models and enable uncertainty analysis of the optimized rate coefficients. ## Large-Scale Virtual Design and Scaffold Diversity Analysis of Triphenylamine Organic Dyes Through Fragment-Based Enumeration and Network-Centric Chemical Space Exploration. - Source: Journal of fluorescence (journals) - Date: 2026-09-07T00:00:00Z - Categories: Cheminformatics - Journal: Journal of fluorescence - DOI: 10.1007/s10895-026-04924-z - External ID: 48c0fafc1e104e956c16b9f845fa271f058d7f7c - Keywords: Chemical Space Exploration - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Machine learning approaches for cancer prognosis and diagnosis via non-coding RNA: a comprehensive review - Source: Briefings in Bioinformatics (journals) - Date: 2026-09-07T00:00:00+00:00 - Authors: Md Shafayet Hossain, Yi-Ping Phoebe Chen - Journal: Briefings in Bioinformatics - DOI: 10.1093/bib/bbag346 - Source URL: - Dashboard article: Abstract: Non-coding RNAs (ncRNAs), once considered genomic dark matter, are now established as key regulators of gene expression with widespread roles in cellular homeostasis and disease. In cancer, ncRNA expression is frequently and systematically dysregulated, and many of these molecules circulate in stable, protected form within biofluids, offering a compelling basis for non-invasive or minimally invasive diagnostic strategies. However, their clinical translation remains substantially hindered to date due to biological complexity, technical noise, and high dimensionality inherent to ncRNA expression datasets. In this context, machine learning (ML) has emerged as a powerful analytical tool to address these challenges, enabling the identification of subtle, reproducible ncRNA signatures predictive of diverse malignancies. This review critically evaluates ML-driven frameworks for cancer diagnosis and prognosis across four ncRNA subclasses, namely miRNAs, lncRNAs, circRNAs, and piRNAs, while also acknowledging the biophysical and thermodynamic models that reinforce ncRNA bioinformatics. Despite substantial methodological progress in ML-based cancer diagnosis and prognosis, key challenges persist, including tumor biological heterogeneity, limited multicenter validation, and the lack of widely adopted standardized protocols for preprocessing, normalization, and reporting workflows. Furthermore, many current ML models lack interpretability in biological or clinical context, constraining their translational utility. By synthesizing recent advances and identifying unresolved barriers, this review charts a roadmap for developing a robust, clinically actionable ncRNA biomarker platform for cancer detection. With global cancer incidence projected to exceed 35 million annual cases by 2050, validated ncRNA-ML-driven frameworks hold potential to revolutionize early-stage detection and personalized therapeutic strategies, thereby reducing the escalating socio-economic burden of cancer worldwide. ## Machine Learning for High-Throughput Reaction Yield Prediction in DNA-Encoded Library Synthesis - Source: ChemRxiv (preprints) - Date: 2026-09-07T00:00:00Z - Categories: Property Prediction, Reaction Informatics, Library Design - Authors: Tao Tang, Li Gao, Sen Gao, Hongyao Zhu, Guansai Liu, Jin Li, Xuemin Cheng - DOI: 10.26434/chemrxiv.15008275/v2 - External ID: 10.26434/chemrxiv.15008275/v2 - Keywords: Reaction Yield Prediction, graph neural network, GNN, DNA Encoded Library, molecular representation, reaction type - Source URL: - Dashboard article: Abstract: DNA-encoded library (DEL) synthesis relies on efficient DNA-compatible reactions to ensure library fidelity and the reliability of downstream affinity selection. However, most existing reaction yield prediction models were developed for general organic synthesis and typically use full-atom representations of complete reactants and products. In DEL single-cycle synthesis validation, an appropriate model reaction is to focus on the exposed reactive handle, its local chemical environment, the incoming building block (BB), and the reaction type, rather than large constant backgrounds such as the DNA tag, linker, and pre-existing scaffold. Here, we propose a functional-group-centric machine learning framework for DEL single-cycle reaction yield prediction. The model uses only the exposed substrate functional group (FG) and its coarse-grained local environment, the incoming BB, and the reaction type as inputs. Benchmarking on a real-world high-throughput DEL single-cycle validation dataset shows that graph neural network (GNN) models based on this localized representation outperform traditional fingerprint-based baselines, with a graph attention network (GAT) achieving the best overall performance and showing strong robustness on highly imbalanced industrial data. These results indicate that foregrounding the local reaction center is not merely a simplification of molecular representation, but a task-aligned strategy that better reflects the chemistry and decision logic of DEL single-cycle transformations. This lightweight framework is readily compatible with practical DEL workflows and provides valuable supports on reagent quality control (QC), building block selection, high-fidelity library design, and retrospective analysis of anomalous results. ## Machine‐Learning‐Decoded Dual‐Channel Photoelectrochemical Fingerprinting at Porphyrinic MOF‐on‐MOF Heterointerfaces for Intelligent Molecular Profiling - Source: Advanced Functional Materials (journals) - Date: 2026-09-07T00:00:00Z - Journal: Advanced Functional Materials - DOI: 10.1002/adfm.78304 - External ID: 5ebaf476be7b77b459a73970f9cf02ce4e319b24 - Source URL: - Dashboard article: Abstract: Accurate discrimination of structurally analogous sulfur‐containing molecules with overlapping photoelectrochemical (PEC) responses remains a major challenge for conventional PEC sensors that rely on one‐dimensional photocurrent readouts. Herein, we develop a dual‐channel PEC fingerprinting platform that pairs Zn‐TCPP with a rationally constructed Cu‐TCPP@Zn‐TCPP MOF‐on‐MOF heterostructure and integrates their paired responses with machine‐learning (ML) algorithms. Unlike traditional single‐signal approaches, Zn‐TCPP and the layered Cu‐TCPP@Zn‐TCPP heterointerface serve as two distinct transduction interfaces that convert molecule‐dependent interfacial interactions into dual‐interface PEC fingerprints. Experimental and theoretical analyses reveal that these discriminative responses arise from coupled molecular and interfacial effects. These effects include molecular‐size‐dependent interfacial accessibility, frontier‐orbital‐influenced hole consumption, and, in the case of glutathione (GSH), Cu–S‐related modulation of the cathodic oxygen‐reduction‐related electron‐consumption pathway. These differences in molecular–interface interactions are used to establish a paired PEC fingerprint dataset. Subsequent ML‐assisted decoding, particularly with an artificial neural network (ANN), resolves the overlapping response patterns. The platform achieves high‐accuracy classification of five representative sulfur‐containing molecules, discrimination among predefined GSH concentration levels, and recognition of binary mixtures and spiked real‐water matrices. This work integrates MOF‐on‐MOF heterointerfacial chemistry with data‐driven fingerprint decoding, providing an interface‐design strategy for intelligent recognition of sulfur‐containing molecules in complex aqueous environments. ## MarkitS: an image-to-SMILES parsing workflow for Markush structures - Source: Journal of Cheminformatics (journals) - Date: 2026-09-07T00:00:00+00:00 - Categories: Cheminformatics - Authors: Kun-Lin Tsai, Yi-Chen Lin, Feng-Chi Chen - Journal: Journal of Cheminformatics - DOI: 10.1186/s13321-026-01301-7 - Keywords: SMILES, Markush structures - Source URL: - Dashboard article: - Abstract: not stored for this record. ## MG-CMIF: A Multi-Granularity Enhanced Cross-Modal Information Fusion Framework for Molecular Property Prediction - Source: Journal of Machine Learning and Information Security (journals) - Date: 2026-09-07T00:00:00Z - Categories: Property Prediction - Journal: Journal of Machine Learning and Information Security - DOI: 10.53941/jmlis.2026.100018 - External ID: e2073caae7b7157c90533f5052e7c1d9dab76ffb - Keywords: Property Prediction, Molecular Property - Source URL: - Dashboard article: Abstract: Molecular property prediction provides an important computational basis for compound screening and drug development by estimating physicochemical characteristics and biological activities from molecular structures. Although deep learning has improved molecular modeling, existing methods often describe molecules through a limited structural view or combine multiple views without sufficiently exploiting their complementary relationships. In addition, graph-based approaches commonly concentrate on local atomic connectivity, making it difficult to represent chemically meaningful structural units that may strongly influence molecular properties. This paper develops MG-CMIF, a multi-granularity cross-modal framework for molecular property prediction. The proposed model describes each molecule from symbolic, topological, and spatial perspectives and learns an integrated representation through three key designs. First, hierarchical graph modeling combines detailed atomic interactions with substructure-level chemical patterns to enrich topology-oriented features. Second, interaction across molecular views enables information relevant to property prediction to be exchanged selectively rather than merged through shallow operations. Third, alignment-oriented training objectives encourage representations derived from the same molecule to preserve compatible chemical semantics during fusion. Experiments on multiple public benchmark datasets show that MG-CMIF achieves better prediction results than competitive methods in both classification and regression settings. Further ablation analyses confirm that hierarchical structural modeling and cross-view integration both contribute to the effectiveness of the proposed framework. ## Multi-Scale No-Pooling Network for Geographical Origin Identification of Boletus via SERS - Source: Analytical Chemistry (journals) - Date: 2026-09-07T00:00:00Z - Categories: Cheminformatics, Property Prediction - Journal: Analytical Chemistry - DOI: 10.1021/acs.analchem.6c05282 - External ID: 1fb84204f5cde0ac15403a41e00cec151bb3e76f - Keywords: molecular fingerprint - Source URL: - Dashboard article: Abstract: Frequent poisoning incidents associated with Boletus mushroom highlight the urgent need for rapid trace detection and precise geographical traceability to ensure public safety. However, traditional spectral analysis faces severe bottlenecks due to the strong interference of complex biological matrices and the highly overlapped spectral characteristics among different origins. To overcome these challenges, this study proposes a collaborative strategy combining surface-enhanced Raman spectroscopy (SERS) technology and a deep learning framework. Using gold nanorods (Au NRs) as the SERS enhancement substrate, the ultra-sensitive detection of the Boletus toxin protein bolesatine peptide was achieved, and key molecular fingerprint information such as tryptophan (733 cm–1) and tyrosine (634 cm–1) could be effectively detected from complex matrices. Subsequently, a multi-scale no-pooling network (MNNet) was constructed, innovatively adopting a dual-stream parallel architecture, which can simultaneously encode microscopic fingerprint details and macroscopic spectral contour trends. The model directly introduced the “no-pooling” mechanism, discarding the downsampling operation in traditional convolutional networks, thereby strictly preserving the spectral-index-to-wavenumber correspondence of Raman shifts and effectively avoiding the loss of key chemical structure information. Under the current internal evaluation setting, the MNNet–SVM framework achieved an accuracy of 97.69% in classifying Boletus samples from four geographical origins and outperformed the tested one-dimensional comparison frameworks. SHAP interpretation and spectral-window masking further indicated that the classification relied on distributed spectral information rather than exclusively on the dominant bolesatine-associated bands. This study successfully established a promising SERS-assisted multivariate analytical strategy for the detection of toxic components and origin identification in complex public safety systems. ## Multifunctional Antidiabetic Effects of Myricetin via α-Glucosidase Inhibition and Enhanced Glucose Utilization - Source: Journal of Integrative and Translational Biomedicine (journals) - Date: 2026-09-07T00:00:00Z - Categories: Docking & Screening - Journal: Journal of Integrative and Translational Biomedicine - DOI: 10.67316/jintb.v1.i3.31 - External ID: c90ae1aa332e2ce4d063eedf868657334d08da0b - Keywords: Molecular docking, binding affinity, EC50 - Source URL: - Dashboard article: Abstract: Background: Diabetes mellitus is also characterized by uncontrolled postprandial hyperglycemia, which is mediated by α-glucosidase activity. Current α-glucosidase inhibitors, e.g., acarbose, are limited by gastrointestinal side effects and localized intestinal action. Myricetin is a natural flavonoid with reported metabolic regulatory properties. This study explores the α-glucosidase inhibitory potential of myricetin and its potential in regulating diabetes-related biological pathways using integrated in silico and in vitro approaches. Methods: Molecular docking was performed to evaluate the binding affinity and interaction of myricetin with α-glucosidase (PDB: 9GSW) in comparison with acarbose. In vitro α-glucosidase inhibitory activity was determined using a colorimetric assay, and inhibitory concentration (IC₅₀) values were calculated. Glucose uptake assay was performed in cells, and effective concentration (EC50) values were calculated. Target prediction followed by functional enrichment analysis was conducted to identify biological processes and signaling pathways regulated by myricetin. Results: Myricetin exhibited stable binding with α-glucosidase (−7.5 kcal/mol), comparable to acarbose (−7.2 kcal/mol). In vitro assays showed concentration-dependent inhibition with an IC₅₀ of 69.42 ± 1.13 µg/mL, while acarbose showed stronger potency (IC₅₀ = 50.71 ± 0.78 µg/mL; p < 0.0001). The glucose uptake assay showed an EC50 of myricetin 59.88 ± 1.00 µg/mL vs metformin = 46.57 ± 2.41 µg/mL; p<0.05. Enrichment analysis revealed that myricetin targets are associated with oxidative stress, inflammation, and metabolic regulation, including PI3K-Akt, MAPK, AGE-RAGE, IL-17, p53, and insulin signaling pathways. Conclusion: Myricetin demonstrates significant α-glucosidase inhibitory activity and potential modulation of key diabetes-related pathways following glucose uptake. Although less potent than acarbose, its multifunctional profile and natural origin highlight its promise as a complementary antidiabetic candidate. ## Navigating Large Chemical Spaces Using Graph Theory and Integer Programming - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-07T00:00:00+00:00 - Categories: Cheminformatics - Authors: Ugochukwu M. Ikegwu, Reid C. Van Lehn, Victor M. Zavala - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c01810 - Source URL: - Dashboard article: Abstract: Navigating and analyzing large chemical spaces are necessary to accelerate the design and discovery of new molecules and chemical processes. In this work, we introduce a computational framework that integrates graph theory and integer programming to enable the efficient navigation of large chemical spaces. Our framework represents the chemical space as a graph, wherein nodes represent molecules and edges represent the degree of similarity or connectivity based on domain-specific information. Using the graph representation, we identify representative molecules by computing the so-called minimum dominating set (MDS), which in our context is the minimum set of molecules that is connected to all other molecules. We present a suite of solution strategies for the MDS problem including heuristic and rigorous integer programming (IP) approaches. We show that these approaches allow us to capture physicochemical properties and domain-specific logic and constraints, facilitating the identification of molecules with the target properties. We demonstrate the effectiveness of the proposed approach by navigating the chemical space of per- and polyfluoroalkyl substances (PFAS); this comprises approximately 15,000 molecular structures. We compare our framework against traditional dimensionality reduction and clustering methods such as t-SNE and K-means clustering. ## Navigating Large Chemical Spaces Using Graph Theory and Integer Programming - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-07T00:00:00Z - Categories: Cheminformatics - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c01810 - External ID: 5592b472c60a4e0164dbd4a0a0afa41b6f792c16 - Source URL: - Dashboard article: Abstract: Navigating and analyzing large chemical spaces are necessary to accelerate the design and discovery of new molecules and chemical processes. In this work, we introduce a computational framework that integrates graph theory and integer programming to enable the efficient navigation of large chemical spaces. Our framework represents the chemical space as a graph, wherein nodes represent molecules and edges represent the degree of similarity or connectivity based on domain-specific information. Using the graph representation, we identify representative molecules by computing the so-called minimum dominating set (MDS), which in our context is the minimum set of molecules that is connected to all other molecules. We present a suite of solution strategies for the MDS problem including heuristic and rigorous integer programming (IP) approaches. We show that these approaches allow us to capture physicochemical properties and domain-specific logic and constraints, facilitating the identification of molecules with the target properties. We demonstrate the effectiveness of the proposed approach by navigating the chemical space of per- and polyfluoroalkyl substances (PFAS); this comprises approximately 15,000 molecular structures. We compare our framework against traditional dimensionality reduction and clustering methods such as t-SNE and K-means clustering. ## Near-Complete Synthon Library via Retrosynthetic Analysis for Automated Organic Synthesis - Source: ChemRxiv (preprints) - Date: 2026-09-07T00:00:00Z - Categories: Reaction Informatics - Authors: Baicheng Zhang, Jing Chen, Xiaolong Zhang, Pieter E. S. Smith, Aoyuan Cheng, Runyang Miao, Hongping Liu, Linjiang Chen, Bin Jiang, Guoqing Zhang, Yi Luo, Jun Jiang - DOI: 10.26434/chemrxiv.15008417/v1 - External ID: 10.26434/chemrxiv.15008417/v1 - Keywords: PubChem, Retrosynthetic, retrosynthesis - Source URL: - Dashboard article: Abstract: As idealized structural fragments for guiding bond disconnections, synthons are considered the conceptual backbone of retrosynthetic analysis. Yet retrosynthesis can be difficult because the choice of synthons requires extensive experience in organic synthesis, which prevents non-experts from accessing desired products. Here we address this knowledge gap by constructing a curated, machine interpretable database of commercially available synthon equivalents (hereafter valid synthons) distilled from large-scale retrosynthetic analysis. Mining over 40 million molecules from PubChem, we identify 1,617 essential valid synthons. These synthons, organized into (i) molecular backbones, (ii) functional groups, and (iii) linker motifs, were obtained via iterative retrosynthetic decomposition and stringent frequency based filtering tied to documented reaction precedents. The resulting set enables the in silico reconstruction of over 94% of these 40 million molecules. We demonstrate that such synthons reduce dependence on expansive reagent inventories, even though the routes themselves are non-conventional and algorithm-generated: using only seven representative valid synthons, we accessed three structurally and functionally divergent targets (melatonin, a natural sleep hormone; an azo dye; and a photoactive melatonin derivative), illustrating the breadth and practicality of a synthons to products methodology in the context of automated synthesis for non-experts in organic synthesis. ## Network and experimental pharmacology to decode the antihyperglycemic action of selected ethnobotanicals - Source: Journal of Integrative and Translational Biomedicine (journals) - Date: 2026-09-07T00:00:00Z - Categories: Docking & Screening - Journal: Journal of Integrative and Translational Biomedicine - DOI: 10.67316/jintb.v1.i3.12 - External ID: c09b75d3c0cf9e2ceafc6c8bbaa62076a0f940f2 - Keywords: Molecular docking, drug likeness, binding affinity, enzyme - Source URL: - Dashboard article: Abstract: Background: Curcuma longa, Emblica officinalis, Momordica charantia, Pterocarpus marsupium, Salacia chinensis, and Tinospora cordifolia are ethnobotanicals with antidiabetic properties. Nevertheless, the effect of these plants when used together against diabetes is unestablished. Therefore, this study investigates the combined action of above-mentioned medicinal plants using network and experimental pharmacology. Methods: In network pharmacology module, compound of all six plants were evaluated for drug-likeness, and their corresponding targets were identified and subjected to enrichment analysis. Additional experiment was conducted using a high-fat diet, low-dose streptozotocin-induced diabetic model. Furthermore, in vitro α-amylase and α-glucosidase enzyme inhibition assays were performed. Results: Among all compounds evaluated, (S)-tembetarine from E. officinalis showed highest drug-likeness score of 1.2. Likewise, enrichment analysis of identified targets revealed distinct cellular components, biological processes, and molecular functions. Molecular docking further showed that Lobatoside I from M. charantia exhibited strongest binding affinity toward α-amylase, with a docking score of -15.587, while (−)-(3S)-1-(3,4-dihydroxyphenyl)-7-(4-hydroxyphenyl)-(6E)-6-hepten-3-ol from C. longa showed strongest binding affinity toward α-glucosidase, with a score of -9.284. Subsequent experimental pharmacology demonstrated that combined extract of these ethnobotanicals effectively reduced hyperglycemia by restoring glucose homeostasis, as confirmed by oral glucose tolerance test and insulin tolerance test. Combined extract also improved lipid metabolism and glycogenesis. Moreover, α-amylase and α-glucosidase enzymes were inhibited by extract in concentration-dependent. Bodyweight showed a gradual increase alongside a reduction in food intake. Additionally, after 18 days of treatment, there was decline in nitrogen metabolites, ameliorated lipid profile, and hepatic glycogen content. Conclusion: This study established notable antidiabetic effects of selected ethnobotanicals through improved glucose homeostasis, inhibition of α-amylase and α-glucosidase enzymes, and amelioration of metabolic abnormalities via multi-target mechanisms. ## Phytochemical Profiling and α-Glucosidase and α-Amylase Inhibitory Activities of Underutilised Nanhaia speciosa Aerial Biomass - Source: Foods (journals) - Date: 2026-09-07T00:00:00Z - Categories: Docking & Screening, Free Energy & MD - Journal: Foods - DOI: 10.3390/foods15173167 - External ID: de11da7bf5b6ca7b9c9b840dc924f3476ebc9a32 - Keywords: Molecular docking, IC50, receptor, molecular dynamics - Source URL: - Dashboard article: Abstract: The aerial parts of Nanhaia speciosa are often discarded despite their potential as a source of bioactive compounds. This study aimed to characterise this underutilised biomass and evaluate its functional properties relevant to carbohydrate digestion. Ultra-high-performance liquid chromatography–quadrupole-Orbitrap high-resolution mass spectrometry (UHPLC-Q-Orbitrap HRMS), Global Natural Products Social (GNPS) molecular networking, and MS/MS fragmentation analysis tentatively annotated 96 compounds, mainly flavonoids and nitrogen-containing constituents. The extract inhibited α-glucosidase and α-amylase in a concentration-dependent manner, with IC50 values of 9.98 ± 0.02 and 174.83 ± 1.31 μg/mL, respectively, compared with 29.73 ± 0.04 and 39.46 ± 0.19 μg/mL for acarbose. The extract also showed measurable chemical antioxidant capacity. Network pharmacology suggested potential involvement of cAMP-, calcium-, and receptor-associated signalling. Molecular docking prioritised butein as a candidate constituent with favourable predicted interactions with both enzymes, and molecular dynamics simulations supported stable predicted binding. The estimated MM/GBSA binding free energies of butein were −36.68 and −32.03 kcal/mol for α-glucosidase and α-amylase, respectively. These findings indicate that the aerial biomass of N. speciosa is a promising source of phytochemicals with potential carbohydrate-hydrolase inhibitory properties, although the activity of individual constituents requires further experimental validation. ## Pretrained 3D Molecular Representations Enable Data-Efficient Discovery of High-Energy-Density Fuels - Source: Chemistry (journals) - Date: 2026-09-07T00:00:00Z - Categories: Cheminformatics, Property Prediction - Journal: Chemistry - DOI: 10.3390/chemistry8090123 - External ID: f56b29034e6ec7988b6e5d3b4d7f099014d14b07 - Keywords: Uni Mol, Molecular Representations, molecular representation - Source URL: - Dashboard article: Abstract: High-energy-density hydrocarbon (HEDH) fuels are essential for aerospace propulsion, yet the design of such fuels is limited by the scarcity of reliable property data. In this work, we fine-tuned Uni-Mol, a pretrained three-dimensional (3D) molecular representation learning framework, on a dataset of 316,069 hydrocarbons from GDB-13. Each molecule in the dataset was labeled with six physicochemical properties calculated using the group contribution method. With only 1% of the training set, Uni-Mol achieved a flash-point mean absolute error (MAE) of 1.67 K. After further training on hydrocarbons with a broader carbon-number range, the model achieved coefficients of determination (R2) of 0.9672–0.9998 across six GDB-17 properties. High-throughput screening of this subset identified seven polycyclic candidates with exceptional energy density and thermal stability. These results demonstrate the potential of pretrained 3D molecular representations for data-efficient molecular discovery and provide a scalable framework for the accelerated identification of next-generation energetic materials. This work provides a foundation for future experimental validation. ## Probabilistic dynamics operators for path-entropy-guided nonequilibrium control of self-assembly - Source: ChemRxiv (preprints) - Date: 2026-09-07T00:00:00Z - Authors: Sanjib Paul, Anna Wojnar, Xun Tang, Alexander J. Pak - DOI: 10.26434/chemrxiv.15008400/v1 - External ID: 10.26434/chemrxiv.15008400/v1 - Keywords: reinforcement learning, molecular dynamics, MD simulations - Source URL: - Dashboard article: Abstract: Self-assembly is inherently dynamical, with competing pathways, metastable intermediates, and kinetic traps influencing which structures ultimately form. The resulting dynamics can be irreversible and generate entropy, providing potentially useful information about how a system progresses toward an assembled state. However, data-driven approaches for controlling self-assembly commonly optimize structural observables that provide limited information about these nonequilibrium dynamics. Here, we introduce a probabilistic dynamics operator (PDO) that learns the time evolution of probability distributions in a reduced collective variable space. The learned operator provides both an efficient dynamical surrogate compared to molecular dynamics (MD) simulations and access to probability fluxes from which entropy production can be quantified. We validate the PDO using discrete Markov ring and continuous triple-well systems and show that models trained exclusively on fixed-temperature trajectories also predict dynamics and accumulated entropy production under unseen, time-varying temperature protocols. We then apply the PDO to patchy particles that competitively assemble into rhombohedral () and trihexagonal morphologies, using the PDO as a surrogate environment for reinforcement learning of temperature protocols. A -morphology-only objective produces yields of 62% comparable to the best fixed-temperature MD simulations, whereas augmenting the objective with a moderate contribution from excess entropy production substantially increases formation to 90% in explicit MD simulations. The optimal temperature control policy exploits transient melting and reorganization before promoting renewed growth, while excessive weighting of entropy production suppresses assembly. These results demonstrate that path entropy-dependent thermodynamic information can provide a useful signal for controlling self-assembly when productive pathways require nonequilibrium reorganization. ## QSARmil: A Package for Molecular Multi-Instance Machine Learning - Source: ChemRxiv (preprints) - Date: 2026-09-07T00:00:00Z - Categories: Cheminformatics, Property Prediction - Authors: Dmitry Zankov, Tomasz Danel, Martin Šícho, Wim Dehaen, Pavel Polishchuk, Mario Barbatti - DOI: 10.26434/chemrxiv.15008449/v1 - External ID: 10.26434/chemrxiv.15008449/v1 - Keywords: RDKit, property prediction, graph neural networks, molecular property - Source URL: - Dashboard article: Abstract: Numerous Python tools are available for molecular machine learning, typically built around core modeling paradigms such as descriptor-based approaches (“RDKit + scikit-learn”), deep neural networks, graph neural networks, geometric deep learning, and foundation models. Multi-instance learning (MIL) was originally inspired by the drug discovery problem, where a molecule can adopt multiple conformers and the biologically active one is unknown. Despite its clear relevance, MIL has not yet been supported by a dedicated and comprehensive Python platform for molecular applications. Here, we present QSARmil, a Python package designed for molecular multi-instance machine learning, that provides an integrated workflow including instance generation (e.g., conformers), 2D and 3D descriptor calculation, implementation of diverse MIL algorithms, and key instance identification. By bringing these components together in a unified and user-friendly framework, QSARmil makes it straightforward to apply multi-instance learning to any molecular property prediction task. ## Quantum computing in drug discovery - Source: Frontiers in Drug Discovery (journals) - Date: 2026-09-07T00:00:00Z - Categories: Design de novo - Journal: Frontiers in Drug Discovery - DOI: 10.3389/fddsv.2026.1815176 - External ID: 21ec14d73a7afdee4fe63a12c55680076dc21297 - Keywords: navigate chemical space - Source URL: - Dashboard article: Abstract: Early-stage decisions in the pharmaceutical development process carry outsized consequences for eventual clinical success, yet the computational tools underpinning these decisions remain fundamentally constrained. Computer-aided drug design (CADD) has transformed how researchers navigate chemical space and predict ligand–target interactions, but classical implementations rely on mechanical force-field approximations that fail to capture polarisation, charge transfer, and electron correlation effects central to molecular recognition and reactivity. Quantum-mechanical treatments that correctly describe these phenomena scale exponentially with system size on classical hardware, rendering them impractical for drug-relevant biomolecules. Quantum computing offers a physically motivated path beyond this scaling barrier: by exploiting superposition, entanglement, and interference, quantum algorithms can in principle simulate electronic structure with polynomial resource requirements for targeted problem classes. This article provides a theoretical review of how quantum computation integrates throughout the drug discovery pipeline, from target identification to lead optimisation. Its primary contribution is a pipeline-level mapping of quantum methods—including the Variational Quantum Eigensolver (VQE), quantum machine learning, and quantum-enhanced optimisation—to specific drug development stages. The review critically distinguishes near-term NISQ (Noisy Intermediate-Scale Quantum: current devices with 50–1,000 noisy qubits operating without full error correction) capabilities from fault-tolerant quantum computing (FTQC) requirements, quantifies current resource gaps through a worked CYP3A4 case study, and identifies algorithmic limitations (barren plateaus, ansatz expressibility, measurement overhead), hardware scalability, and error correction overhead as the principal barriers to practical deployment. ## Raman and AFM-IR Spectroscopy in the Characterization of Giant Plasma Membrane Vesicles Isolated from Microglia and Glioblastoma Cells - Source: Langmuir (journals) - Date: 2026-09-07T00:00:00Z - Categories: Cheminformatics, Spectra & Analytical - Journal: Langmuir - DOI: 10.1021/acs.langmuir.6c02670 - External ID: 9874187d1cf1100893e8cf44c0f2de18affbf723 - Keywords: molecular fingerprints, Chemometric - Source URL: - Dashboard article: Abstract: Giant plasma membrane vesicles (GPMVs) are widely used as cell-derived models of biological membranes, yet their spectroscopic characterization remains limited. In this work, we propose a preparation and analytical workflow enabling Raman and Atomic Force Microscopy-based infrared (AFM-IR) spectroscopic investigation of GPMVs isolated from microglial (human microglial clone 3, HMC3) and glioblastoma (LN229, U87 MG, and U251 MG) cell lines. Storage conditions and isolation parameters were systematically evaluated to ensure preservation of vesicle morphology and biochemical stability prior to spectroscopic measurements. Raman spectroscopy provided label-free molecular fingerprints of GPMV composition, revealing variations in lipid and protein contributions depending on the cellular origin of the vesicles. Complementary characterization was achieved using AFM-IR spectroscopy, enabling the detection of differences in carbohydrate-, protein-, and lipid-related vibrational bands at the nanoscale, without the need for sample labeling. Chemometric analysis, including principal component analysis and partial least squares discriminant analysis, further confirmed the spectroscopic discrimination of vesicles derived from different cell lines. To our knowledge, this work represents one of the first combined Raman and AFM-IR spectroscopic investigations of GPMVs, demonstrating the potential of multimodal vibrational spectroscopy for the characterization of cell-derived membrane systems. The presented methodology establishes a foundation for future spectroscopic studies of membrane-derived vesicles and their biochemical heterogeneity. ## Structure based discovery of VMP-7 as a promising GlgE targeting lead against Mycobacterium tuberculosis - Source: Frontiers in Microbiology (journals) - Date: 2026-09-07T00:00:00Z - Categories: Cheminformatics, Property Prediction - Journal: Frontiers in Microbiology - DOI: 10.3389/fmicb.2026.1909418 - External ID: 352616a6c3b0ff90cd21970dbf74a34e620f2b61 - Keywords: molecular fingerprint, virtual screening, enzyme - Source URL: - Dashboard article: Abstract: Tuberculosis (TB), caused by Mycobacterium tuberculosis ( M.tb ), remains a major global health challenge, necessitating the identification of novel drug targets with high specificity and therapeutic potential. Maltosyltransferase (GlgE) is a critical enzyme in the biosynthesis of cytosolic α-glucan, a key component required for maintaining cell wall integrity and intracellular survival of the pathogen. GlgE catalyzes the transfer of maltosyl units from maltose-1-phosphate to elongating α-glucan chains, and its inactivation results in toxic accumulation of metabolic intermediates, ultimately leading to bacterial death. Importantly, the absence of homologous pathways in humans underscores its suitability as a selective anti-tubercular (anti-TB) target. Sequence analysis of 382 clinical isolates demonstrated complete conservation of the glgE gene, highlighting its evolutionary stability and essentiality. Subsequently, structure-based virtual screening followed by molecular fingerprint clustering was performed to identify chemically diverse inhibitors. Twelve top ranked compounds (VMP-1 to VMP-12) were further evaluated through in-vitro antimycobacterial assays in broth culture and infected THP-1 derived macrophages, along with cytotoxicity assessment. Among these, VMP-7 exhibited significant inhibitory activity against M.tb at 12.5 μg/mL, with minimal cytotoxic effects on host cells. Collectively, these findings validate GlgE as a highly conserved target and identify VMP-7 as a promising lead candidate for further preclinical development of novel anti-TB therapeutics. ## Structure‐Guided Identification of GPR84 Ligand Candidates With Potential Immunomodulatory Activities - Source: ChemMedChem (journals) - Date: 2026-09-07T00:00:00+00:00 - Categories: Docking & Screening, ADMET & Safety - Authors: Ha‐Anh Thi Pham, Anamaria Morales‐Alvarez, Timothy Gilbertson, Hung Nguyen - Journal: ChemMedChem - DOI: 10.1002/cmdc.70488 - Keywords: de novo design, molecular docking, receptor, molecular dynamics, ADME - Source URL: - Dashboard article: Abstract: GPR84, a medium‐chain fatty acid (MCFA)‐sensing G protein‐coupled receptor, plays a significant role in immune regulation. Despite its therapeutic potential, structurally diverse and well‐characterized chemical probes of GPR84 remain limited. In the current study, we developed a novel structure‐guided and dual‐route pipeline to explore and prioritize small‐molecule scaffolds compatible with the GPR84 orthosteric pocket, with pharmacophore‐based ZINC screening and LigBuilder‐based de novo design, accompanied by molecular docking, ADME filtering, and molecular dynamics, followed by immunoregulatory activity validation. ZINC screening identified ZINC149375 (SC787) and ZINC76994316 (Z109), which were further evaluated by 500‐ns molecular dynamics simulations. Additionally, LIG081 and LIG133 from the LigBuilder‐derived series emerged as computationally prioritized de novo candidates with favorable and persistent predicted interactions within the GPR84 orthosteric pocket. These ZINC‐derived compounds demonstrated potent immune cell‐dependent immunomodulatory activities associated with GPR84 functional agonism. SC787 and Z109 promote proinflammatory signaling in macrophages while suppressing activation and effector cytokine production in activated CD8 + T cells. Collectively, this study identifies structurally prioritized GPR84 pocket‐compatible scaffolds for future receptor‐focused immunological activity validation and medicinal chemistry optimization. ## Synthesis, DFT Evaluation, and Docking-Based Assessment of Novel Substituted Pyridine Derivatives against Human Spermine Oxidase - Source: Journal of the Turkish Chemical Society Section A: Chemistry (journals) - Date: 2026-09-07T00:00:00Z - Journal: Journal of the Turkish Chemical Society Section A: Chemistry - DOI: 10.18596/jotcsa.1894898 - External ID: 6edfcc546e65c9d8e6b2fc9a183e17578ee753eb - Keywords: molecular docking, DFT, B3LYP, Suzuki Miyaura - Source URL: - Dashboard article: Abstract: This study combines structure-based molecular docking simulations and DFT calculations to investigate interactions between the crystal structure of human Spermine Oxidase (SMOX) and a series of three novel 2,4,6-substituted pyridine compounds as docking-prioritized SMOX-interacting candidates. The synthesis employed tandem iridium-catalyzed borylation and Suzuki-Miyaura coupling. DFT calculations were performed at the B3LYP/6-311(d,p) level to determine the global minimum energy geometries and electronic properties. These optimized structures served as inputs for molecular docking. The interaction potential was simulated using the human SMOX crystal structure (PDB ID: 7OXL). Docking grids were centered on the FAD-binding catalytic domain. The precise coordinates were x = -25, y = 93, and z = 60. These coordinates ensured rigorous active-site definition. The synthesized scaffolds exhibited superior binding affinities with Vina scores ranging from -8.3 to -9.7 kcal/mol. These scores are significantly higher than the natural substrate, spermine (-5.8 kcal/mol). Structural analysis used the FAD cofactor as a landmark, supporting a hypothesized "Channel Plug" mechanism in which the bulky heterocyclic core sterically hinders access to the catalytic center. The results illustrate that the key interacting residues include Ser463, Gly13, Ala36, Glu35, Leu56, Leu14, Ala15, and Thr465. These findings provide a robust theoretical structural rationale for developing site-specific SMOX modulators and prioritize the use of pyridine scaffolds to target polyamine-binding cavities. ## Systematic Evaluation of Graph Neural Networks for Ligand-Based Virtual Screening on ChEMBL Datasets - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-07T00:00:00+00:00 - Categories: Cheminformatics, Property Prediction, Docking & Screening - Authors: Haihan Liu, Jiaqi Lin, Ying Fan, Jintong Du, Xinying Yang, Hao Fang, Xuben Hou - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c02160 - Keywords: Virtual Screening, ChEMBL, Graph Neural Networks, GNNs, GNN - Source URL: - Dashboard article: Abstract: The performance of target-specific, ligand-based virtual screening models is strongly influenced by dataset characteristics, including data availability, class imbalance, and evaluation strategies. In this work, we perform a systematic evaluation of graph neural networks (GNNs) using a ChEMBL-derived dataset spanning 5,368 targets and 1.59 million activity records, capturing the long-tailed distributions and target-specific imbalances commonly observed in pharmaceutical data. Through a systematic evaluation of multiple GNN architectures, we identify guidelines for model selection: while the Graph Isomorphism Network (GIN) consistently outperforms others on datasets with >100 samples (a mean ROC–AUC up to 0.94), simpler architectures are more robust under extreme data scarcity. Critically, our comparative analysis of splitting strategies reveals that random sampling yields artificially optimistic performance due to structural overlaps, whereas similarity-aware clustering exposes a substantial generalization gap (AUC drop > 0.3), cautioning against prevailing evaluation practices. We further demonstrate that multi-task learning serves as an effective remedy for small-target instability, providing significant and consistent performance gains. To underscore its translational value, we deploy this comprehensive framework in a virtual screening campaign against Mcl-1, yielding a chemically optimized lead, C4 (Ki = 0.58 μM), with verified cellular efficacy. Our findings highlight the importance of task-aware benchmark design and offer a practical strategy for reliable GNN application in drug discovery. ## Systematic Evaluation of Graph Neural Networks for Ligand-Based Virtual Screening on ChEMBL Datasets - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-07T00:00:00Z - Categories: Cheminformatics, Property Prediction - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c02160 - External ID: ea4e7dc45d5ddcb42f1ad3961e492563e8f8dde4 - Keywords: Virtual Screening, ChEMBL, Graph Neural Networks, GNNs, GNN - Source URL: - Dashboard article: Abstract: The performance of target-specific, ligand-based virtual screening models is strongly influenced by dataset characteristics, including data availability, class imbalance, and evaluation strategies. In this work, we perform a systematic evaluation of graph neural networks (GNNs) using a ChEMBL-derived dataset spanning 5,368 targets and 1.59 million activity records, capturing the long-tailed distributions and target-specific imbalances commonly observed in pharmaceutical data. Through a systematic evaluation of multiple GNN architectures, we identify guidelines for model selection: while the Graph Isomorphism Network (GIN) consistently outperforms others on datasets with >100 samples (a mean ROC–AUC up to 0.94), simpler architectures are more robust under extreme data scarcity. Critically, our comparative analysis of splitting strategies reveals that random sampling yields artificially optimistic performance due to structural overlaps, whereas similarity-aware clustering exposes a substantial generalization gap (AUC drop > 0.3), cautioning against prevailing evaluation practices. We further demonstrate that multi-task learning serves as an effective remedy for small-target instability, providing significant and consistent performance gains. To underscore its translational value, we deploy this comprehensive framework in a virtual screening campaign against Mcl-1, yielding a chemically optimized lead, C4 (Ki = 0.58 μM), with verified cellular efficacy. Our findings highlight the importance of task-aware benchmark design and offer a practical strategy for reliable GNN application in drug discovery. ## TAME: Element-wise Mixture-of-Experts Fusion for Reliable and Interpretable Molecular Property Prediction - Source: ChemRxiv (preprints) - Date: 2026-09-07T00:00:00Z - Authors: Robert P. Schiller, Christoph Weisser, Klaus-Robert Müller, Christian Ochsenfeld, Parastoo Semnani - DOI: 10.26434/chemrxiv.15008270/v2 - External ID: 10.26434/chemrxiv.15008270/v2 - Keywords: Property Prediction, graph neural networks, Molecular Property, molecular representations - Source URL: - Dashboard article: Abstract: Property prediction for molecules and materials is bottlenecked by label scarcity, and in this regime the operative failure mode is not a low mean but an unpredictable tail: individual training runs collapse. The three dominant molecular representations fail in complementary ways—physicochemical descriptors are exact but fixed, graph neural networks learn topology but degenerate under sparse supervision, and language models supply semantic context but cannot compute exact quantities— yet combining them is itself the learning problem, because modality-level scalar gating commits the whole model to one trust weight per source. We introduce TAME , a tri-modal encoder whose element-wise Mixture-of-Experts router assigns an independent modality mixture to every hidden coordinate, stabilized by a balance–entropy regularizer, a closed-form gate initialization (β = log τ ) that is intended to remove router burn-in, and two-stage self-supervised graph pretraining. On scaffold-split BACE (≈ 1.5k molecules, 100 seeds) we evaluate two fusion topologies, each against its own graph-only control. In both, adding the text and descriptor experts contracts the seed-to-seed distribution of the threshold-free metrics—ROC-AUC σ falls five-fold in the flat topology and 1.6-fold in the hierarchical one, whose control was already the tighter of the two—and the collapse tail reaching ROC-AUC ≈ 0.45 that the weaker single-modality configurations carry is absent from the fused models while accuracy is retained. Each comparison is matched on everything but the fusion stage, isolating its effect. The router allocates a distinct mixture to each coordinate, which no scalar gate can represent, and the entropy weight moves that allocation continuously between graded mixing and a near-binary regime in which each coordinate is claimed outright by one modality. Fused-representation alignment tracks encoder quality without supervision, shifting from text to topology once pretraining lifts the graph embedding out of rank-1 collapse. The same three experts map onto crystal graphs, composition–structure descriptors and literature-mined text. These three modalities recur unchanged across molecular and inorganic chemistry, so TAME is a single interpretable recipe for turning volatile low-data predictors into reliable, self-explaining ones, wherever data are scarce and every failed prediction costs a real experiment. ## The MAZ-POLD1 Signaling Axis Drives Cisplatin Resistance in Bladder Cancer by Activating DNA Damage Repair - Source: Cancers (journals) - Date: 2026-09-07T00:00:00Z - Categories: Docking & Screening - Journal: Cancers - DOI: 10.3390/cancers18172892 - External ID: 2892df49bb3b398d70031c265acf8d6169503f4c - Keywords: molecular docking - Source URL: - Dashboard article: Abstract: Background: Although POLD1 exhibits oncogenic properties in multiple malignancies, its precise role and regulatory mechanisms in cisplatin resistance of bladder cancer (BC) remain elusive. This study aims to elucidate the functional involvement and upstream regulatory axis of POLD1 in BC chemoresistance. Methods: We evaluated the impact of POLD1 on chemoresistance and DNA damage repair (DDR) using public clinical databases and cisplatin-resistant cell lines, employing CCK-8, colony formation, flow cytometry, immunofluorescence, and comet assays. Protein interactions were examined via Co-IP, molecular docking, and truncation mutant analysis. ChIP-PCR and dual-luciferase reporter assays were utilized to identify the upstream transcription factor. In vivo functionality was further validated in a nude mouse xenograft model. Results: POLD1 was markedly upregulated in BC tissues and cisplatin-resistant BC cells, and its elevated expression was tightly associated with advanced tumor stage, high pathological grade, and poor patient prognosis. Functional experiments verified that POLD1 knockdown aggravated cisplatin-induced DNA damage and drastically sensitized BC cells to cisplatin in vitro, and attenuated tumor growth under cisplatin treatment in vivo, indicating that POLD1 is a key driver of cisplatin resistance. Mechanistically, POLD1 physically interacted with and activated ATM, thereby initiating homologous recombination (HR) repair signaling. Moreover, transcription factor MAZ directly bound the promoter region of POLD1 to transcriptionally upregulate its expression. Rescue experiments in vitro further validated that the MAZ-POLD1 axis facilitates cisplatin resistance by modulating the DDR pathway in BC. Conclusions: Therapeutic targeting of the POLD1-regulated DDR pathway holds great promise as an effective strategy to reverse acquired cisplatin resistance in BC. ## Unraveling the hepatic and immune toxicity of triclosan: an integrative network toxicology, molecular docking, and experimental study - Source: BMC Gastroenterology (journals) - Date: 2026-09-07T00:00:00Z - Categories: Docking & Screening - Journal: BMC Gastroenterology - DOI: 10.1186/s12876-026-05296-1 - External ID: 581948cb31147ad45f1abfb4563d2a2cbf64cbb5 - Keywords: molecular docking - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Unraveling tumor cell heterogeneity and epithelial-mesenchymal plasticity in gastric adenocarcinoma: an integrative multi-omics framework evaluating PVR/CD155 as a tumor cell-intrinsic EMT-associated target - Source: Frontiers in Immunology (journals) - Date: 2026-09-07T00:00:00Z - Categories: Docking & Screening, Targets & Structures - Journal: Frontiers in Immunology - DOI: 10.3389/fimmu.2026.1932169 - External ID: 5cbd219109074e8150052cd4114d4cb1fa41a336 - Keywords: molecular docking, molecular dynamics - Source URL: - Dashboard article: Abstract: The tumor microenvironment (TME) of gastric adenocarcinoma is exceptionally heterogeneous, and epithelial-mesenchymal transition (EMT) is a central mechanism promoting local invasion and metastatic spread. Even so, how EMT-programmed cells are spatially arranged within tumor tissue, how they interact with neighboring immune populations, and which molecular nodes might be exploited therapeutically remain incompletely defined. We built a large-scale, multi-layered analytical pipeline that combined single-cell transcriptomics (approximately 250,000 cells drawn from two independent patient cohorts), Visium-based spatial transcriptomics processed through Bayesian cell2location deconvolution, niche-level SpaTopic modeling, deep-learning histology analysis (ResNet50 feature extraction paired with CellProfiler-derived morphometrics), and ensemble survival modeling spanning over one hundred algorithmic combinations. Gaussian mixture modeling was used to define EMT-high cell states, the Scissor framework was applied to link fibroblast subsets to patient mortality, and a multi-tier filtering scheme was used to nominate druggable candidate genes. The top candidate, PVR/CD155, was interrogated experimentally by RT-qPCR, Western blotting, immunohistochemistry, and siRNA knockdown in AGS cells. We further evaluated PVR druggability by molecular docking, a 100-nanosecond molecular dynamics trajectory, and MM/GBSA binding-energy estimation using PP-121 as a candidate ligand. Sub-clustering identified seven fibroblast subclusters: Fib\_APOD, Fib\_COL4A1, Fib\_SLPI, Fib\_COL1A1, Fib\_CCL4, Fib\_STMN1, and Fib\_S100B. The SLPI-high subset preferentially localized to peritoneal metastases. Phenotype-guided Scissor mapping nominated a survival-associated fibroblast program enriched within COL4A1-expressing fibroblast states. PVR/CD155 was prioritized as an EMT-linked candidate gene; single-cell expression profiling showed that PVR was most frequently detected in endothelial, epithelial and fibroblast compartments, with low detection in lymphoid and plasma cells. PVR/CD155 was significantly upregulated in gastric cancer cell lines and tumor tissues. PVR/CD155 knockdown in AGS cells decreased proliferation, migration and invasion, supporting a tumor cell-intrinsic functional role rather than establishing PVR as a stromal immune biomarker. Molecular docking and dynamics simulations identified PP-121 as a candidate PVR-binding ligand for future biochemical validation. The proposed framework connects single-cell resolution with spatial and histological data, produces validated prognostic tools and nominates PVR/CD155 as a tumor cell-intrinsic EMT-associated target candidate for gastric cancer. Stromal or immune regulatory roles of PVR remain plausible but require direct compartment-specific and functional validation. ## uPAR Amplifies Macrophage Inflammation via NF-κB Pathway to Promote Fibrotic Transition Following Acute Kidney Injury. - Source: Antioxidants & redox signaling (journals) - Date: 2026-09-07T00:00:00Z - Categories: Docking & Screening - Journal: Antioxidants & redox signaling - DOI: 10.1177/15230864261486342 - External ID: f9f65bb5f7a40ea0eb62df2e0509d180b7f6f93f - Keywords: Molecular docking, receptor - Source URL: - Dashboard article: Abstract: AIMS Persistent inflammation is recognized as a major driver of the acute kidney injury (AKI) to chronic kidney disease (CKD) transition, yet upstream macrophage-activation signals remain incompletely understood. Here, we investigated whether the urokinase receptor (uPAR), traditionally linked to matrix remodeling, functions instead as an inflammatory signaling hub that links kidney injury to chronic fibrotic remodeling. RESULTS In a unilateral renal ischemia-reperfusion injury model, uPAR was induced during the fibrotic phase of postischemic kidney injury. Functional enhancement- and loss-of-function approaches revealed a striking dichotomy: exogenous urokinase (uPA) aggravated renal dysfunction, inflammation, oxidative stress, and fibrosis, whereas uPAR deletion was protective. Mechanistically, uPA/uPAR signaling amplified macrophage inflammatory responses, enhancing M1 polarization, pro-inflammatory cytokine production, reactive oxygen species generation, and NF-κB activation. Molecular docking, mutational modeling, and co-immunoprecipitation analyses revealed a previously underappreciated association between activated uPAR and toll-like receptor 4 (TLR4)-containing complexes. This interaction enhanced myeloid differentiation primary response gene 88-dependent NF-κB signaling and potentiated macrophage inflammatory amplification rather than initiating inflammation independently. NF-κB blockade abolished the pro-inflammatory effects of uPA/uPAR signaling, establishing the functional importance of this pathway. INNOVATION We demonstrate that uPA-activated uPAR engages TLR4-associated signaling to intensify macrophage-driven inflammation, oxidative stress, and fibrotic remodeling. These findings redefine uPAR as a molecular switch governing maladaptive kidney repair and identify the uPA/uPAR-TLR4 signaling interface as a promising therapeutic target. CONCLUSION Our findings suggest that the uPA/uPAR-TLR4 axis is involved in regulating the progression from AKI to fibrosis. Targeting this signaling node may represent a novel strategy to interrupt maladaptive repair and prevent CKD following AKI. Antioxid. Redox Signal. 00, 000-000. ## XenoSoM: A deep learning framework for site of metabolism prediction of xenobiotics. - Source: European journal of medicinal chemistry (journals) - Date: 2026-09-07T00:00:00Z - Journal: European journal of medicinal chemistry - DOI: 10.1016/j.ejmech.2026.119312 - External ID: a65b8525415165bd3a1c072870be4e421a0129a7 - Keywords: GNN, enzyme, metabolism prediction - Source URL: - Dashboard article: Abstract: Metabolic biotransformations significantly influence drug efficacy and safety, making early metabolism assessment crucial in drug discovery. While in vivo and in vitro methods remain the gold standard for investigating metabolic properties, they are often costly and time-consuming. Consequently, AI-driven models have emerged as useful tools to predict metabolic reactions, sites of metabolism (SoM), and resulting metabolites. Predicting the SoM is particularly valuable for identifying reactive atoms and anticipating compounds' metabolic profile. In this study, six deep learning architectures (GCN, GIN, GATC, GINE, GATv2 and AttentiveFP) were trained on a curated dataset containing more than 30'000 substrates with annotated SoMs, thus addressing the long-standing scarcity of large, high-quality datasets in computational metabolism prediction. The final selected model, named XenoSoM, employs an AttentiveFP architecture with single-task (ST) and multi-task (MT) variants, both delivering the best performance among the evaluated models, achieving MCC values up to 0.848 and Top-2 accuracies up to 1.00. XenoSoM\_ST and XenoSoM\_MT were compared against existing tools such as FAME3R and GNN-SOM. The comparison showed that our architectures perform better than FAME3R and GNN-SOM, representing a notable advancement in in silico drug metabolism modeling for both phase I and phase II reactions. Despite the inclusion of proprietary data sources and the lack of enzyme-specific information, XenoSoM represents an important step toward the development of global models for metabolism prediction, leveraging large-scale data integration to improve the identification of metabolic hotspots. ## XGBoost-Based Prediction of pKa Values from Molecular Descriptors - Source: Journal of Undergraduate Research International (journals) - Date: 2026-09-07T00:00:00Z - Categories: Cheminformatics, Property Prediction, ADMET & Safety - Journal: Journal of Undergraduate Research International - DOI: 10.64589/juri/221642 - External ID: da9e711b60ca6072acf71f2ade0f8c12ee803cec - Keywords: ECFP, MACCS keys, RDKit, XGBoost, Gradient Boosting, logP, dissociation constant, pKa prediction, Molecular Descriptors, molecular descriptor - Source URL: - Dashboard article: Abstract: The acid dissociation constant (pKa) is a fundamental physicochemical property that governs molecular ionization, solubility, and bioavailability, rendering its accurate prediction essential in drug design and environmental toxicology. This study entailed a systematic comparison of molecular descriptor representations for pKa prediction using eXtreme Gradient Boosting (XGBoost) regression. Four descriptor categories were evaluated: Morgan circular fingerprints, known as extended-connectivity fingerprints (ECFP), with varying radii (one to four) and bit lengths (12–16,384), Molecular ACCess System (MACCS) structural keys, physicochemical descriptors (molecular weight, logP, rotatable bond count, and fragment counts), and combined representations. A comprehensive grid search optimized the XGBoost hyperparameters across all configurations using five-fold cross-validation on a curated dataset of 7,912 compounds with experimental pKa values obtained from DataWarrior. Morgan fingerprints with radius two and 16,384 bits achieved the highest standalone performance (test R2 = 0.795, root mean squared error, RMSE = 1.539), whereas combining Morgan fingerprints (radius one, 2,048 bits) with MACCS keys and physicochemical descriptors yielded the optimal overall result (test R2 = 0.817; RMSE = 1.455). An extended analysis employing 1,135 Mordred and RDKit cheminformatic descriptors with XGBoost feature selection further improved the prediction accuracy (test R2 = 0.829; RMSE = 1.404). Moreover, feature importance analysis identified the number of acidic groups, spectral descriptors, and aromatic amine fragments as the most influential predictors. These results demonstrate that multisource descriptor fusion consistently outperforms individual representations and that XGBoost provides a computationally efficient and interpretable framework for pKa modeling. ## DrugReason: Dynamic Multi-View Reasoning over Knowledge Graph and Language Evidence for Drug Repurposing - Source: arXiv (preprints) - Date: 2026-09-06T18:54:26Z - Categories: Property Prediction, LLMs & Agents - Authors: Zijie Liu, Hongxuan Li, Zhen Tan, Jinhao Duan, Baixiang Huang, Zunpeng Liu, Kai Shu, Tianlong Chen - External ID: 2609.06779v1 - Keywords: LLM, LLMs - Source URL: - Dashboard article: - PDF: Abstract: Drug repurposing aims to identify new therapeutic uses for existing compounds and, compared with de novo drug discovery, offers a faster and more cost-effective path to clinical translation. However, the space of candidate drug-disease pairs is enormous and their underlying relationships often depend on complex multi-hop biological mechanisms, making it difficult to reliably predict which pairs represent true therapeutic relationships. Existing approaches tackle this from two directions: knowledge graph-based methods organize curated biomedical evidence into structured relational networks for grounded multi-hop reasoning, while LLM-based methods leverage pretrained knowledge to generate flexible mechanistic rationales. Yet neither is sufficient alone - KGs are confined to observed graph structure while LLMs lack factual grounding and risk hallucination. To address this gap, we propose DrugReason, a multi-view reasoning framework that integrates grounded KG reasoning with LLM-generated mechanistic inference for drug repurposing. DrugReason adaptively routes diverse reasoning paths to specialized experts conditioned on the query context, while a cross-expert distillation objective enables knowledge sharing without sacrificing expert specialization. Experiments on PharmaDB, DDInter, and DrugBank show that DrugReason improves average performance over strong single-view reasoning baselines and achieves competitive or superior results compared with graph-based alternatives, while providing interpretable routing-based predictions. ## Out-of-Distribution Inverse Design of Elastic Networks with Differentiable Graph Neural Network Molecular Dynamics - Source: arXiv (preprints) - Date: 2026-09-06T15:09:05Z - Categories: Property Prediction, Design de novo - Authors: Sergey A. Shteingolts, Salman N. Salman, Ron Levie, Dan Mendels - External ID: 2609.06655v1 - Keywords: Inverse Design, Graph Neural Network, Molecular Dynamics - Source URL: - Dashboard article: - PDF: Abstract: Machine-learning-based inverse design can accelerate the discovery of materials with targeted properties, but conventional structure--property models often require large training datasets and generalize poorly beyond their training distribution. Here, we present a differentiable inverse design framework based on a graph neural network molecular dynamics simulator. By combining a short dynamical initialization with physics-based refinement during simulation, the framework enables optimization well beyond the conditions represented in the training data. Using disordered elastic networks, we show that a simulator trained only on non-auxetic systems with Poisson's ratios between 0.1 and 0.4 can design strongly auxetic networks with values as low as -0.3. The framework also produces stress--strain responses outside the range observed during training, creates localized mechanical defects absent from the training data, and generalizes across system size, enabling optimization of networks tested up to 5000 nodes despite being trained only on networks with fewer than 200 nodes. ## A drug design themed escape room approach to enhance pharmacy resident learning. - Source: Currents in pharmacy teaching & learning (journals) - Date: 2026-09-06T00:00:00Z - Journal: Currents in pharmacy teaching & learning - DOI: 10.1016/j.cptl.2026.102782 - External ID: f2ab40b10a2d5eec97c28d826741f62e5894f239 - Keywords: toxicity prediction - Source URL: - Dashboard article: Abstract: OBJECTIVE Escape room-based learning has shown promising outcomes in pharmacy education. This study aimed to describe the design and implementation of an educational escape room focused on drug design and to explore residents' knowledge improvement through pre- and post-test assessments, as well as their perceptions of the activity. METHODS The educational escape room was developed and tested with 20 pharmacy residents who were divided into four groups of five. The escape room consisted of six sequential puzzles that addressed the principles of drug design, druglikeness rules, toxicity prediction, and biological targets, leading to an in silico study task. The session time was set at 55 min. A pre-test and post-test assessed knowledge acquisition, and a satisfaction questionnaire evaluated residents' perceptions. Pre- and post-test scores were compared using a Wilcoxon signed-rank test, and effect size was estimated using Cohen's dav for paired data. RESULTS All groups successfully completed the challenge, with a mean completion time of 50 min. Scores increased from pre-test to post-test (4.85 ± 1.31/10 vs. 6.80 ± 1.32/10; p < 0.001), with a large effect size (Cohen's dav = 1.48). Most of the residents were satisfied with the escape room, saying that they would recommend it to their colleagues. They reported positive perceptions of the escape room, particularly regarding its contribution to knowledge acquisition and its engaging and stimulating nature. CONCLUSION In general, the escape room approach aimed to reinforce residents' understanding of drug design concepts and in silico methodologies, promote active learning, suggesting its potential for broader application in pharmacy education. ## Advancing Fluorine Mass Balance Using Machine Learning-Based Quantitative Nontarget Screening - Source: ChemRxiv (preprints) - Date: 2026-09-06T00:00:00Z - Authors: Jonathan Zweigle, Merle Plassmann, Thomas M. Karlsson, Emma L. Schymanski, Jonathan P. Benskin, Jan H. Christensen - DOI: 10.26434/chemrxiv.15008377/v1 - External ID: 10.26434/chemrxiv.15008377/v1 - Keywords: Gradient Boosting, XGBoost - Source URL: - Dashboard article: Abstract: Per- and polyfluoroalkyl substances (PFAS) often account for only a small fraction of the extractable organic fluorine (EOF) in fluorine mass balance (FMB) studies. Closing this fluorine gap requires identification and quantification of a broader range of fluorinated compounds, including low-fluorinated substances that are difficult to prioritize by nontarget screening (NTS). Here, a workflow combining fluorine-specific NTS with machine learning-based quantification (qNTS) to improve FMB closure was developed. Extreme Gradient Boosting (XGBoost)-based qNTS models for electrospray ionization (ESI) + and ESI − were trained with 597 reference compounds (185 fluorinated) reaching mean absolute error of ×2.8 and ×2.5 on held-out folds, respectively. An independent validation set of 57 compounds previously unseen by the model reached a mean error of ×2.1. The workflow was applied to urban runoff extracts and AFFF-impacted soils. In sewage-associated runoff, fluorinated pharmaceuticals and their transformation products dominated the identified fluorine, whereas conventional PFAS (i.e. those containing a perfluoroalkyl chain) contributed <3%. Non-PFAS organofluorines (including many pharmaceuticals) contributed up to 48% of the EOF. Identified compounds explained 54–107% of measured EOF in sewage-associated samples but only 1–24% in street runoff, indicating a large fraction of unidentified fluorine in the latter. Application to aqueous film forming foam (AFFF)-impacted soils showed that qNTS performed well for anionic PFAS but was less reliable for cationic and zwitterionic PFAS, which were poorly represented in the training data. Overall, qNTS provided concentration estimates that enabled rapid identification of the compounds contributing most to EOF, thereby directing subsequent target confirmation and FMB closure. ## AHR‐Responsive Carbon Dots Orchestrate Skin Wound Healing and Colonic Mucosal Repair via Treg‐Mediated Macrophage Efferocytosis - Source: Advanced Science (journals) - Date: 2026-09-06T00:00:00Z - Categories: Docking & Screening - Journal: Advanced Science - DOI: 10.1002/advs.77526 - External ID: 1f7a5baea95101643343c7a0170d358419623eed - Keywords: pharmacophore modeling, receptor - Source URL: - Dashboard article: Abstract: As critical interface organs, the gut and skin share substantial immunological similarities, and gut diseases are often accompanied by skin comorbidities, yet the underlying mechanisms remain unclear. Here, we observed delayed skin wound healing in a mouse model of colitis. To promote both skin and intestinal repair, we designed aryl hydrocarbon receptor (AHR)–responsive carbon dots using pharmacophore modeling and fragment‐based drug design. These nanoscale carbon dots significantly alleviated colitis and accelerated impaired skin wound healing. The carbon dots preferentially accumulated in the inflamed colon, where they induced infiltration of tissue‐resident regulatory T (Treg) cells, thereby suppressing mucosal inflammation and promoting tissue repair. Mechanistically, the carbon dots activated AHR signaling to drive Treg cell differentiation. Importantly, Treg‐derived chemokine CCL1 enhanced macrophage efferocytosis, establishing a pro‐regenerative immune microenvironment in both the colon and skin wounds via STAT3–SCARB1 signaling. Notably, Ccl1 knockdown attenuated the therapeutic effects of carbon dots on both colonic mucosal repair and skin wound healing, supporting CCL1 as a key mediator of gut–skin immune communication. Collectively, this study highlights the important role of Treg cells in gut–skin axis crosstalk and provides a nanomaterial‐based strategy for coordinated tissue repair. ## An integrative evaluation of in vitro anti-inflammatory potential, network pharmacology and molecular docking of Girardinia diversifolia (Urticaceae) relevant to rheumatoid arthritis - Source: In Silico Pharmacology (journals) - Date: 2026-09-06T00:00:00Z - Categories: Docking & Screening - Journal: In Silico Pharmacology - DOI: 10.1007/s40203-026-00730-4 - External ID: b3178cc841b488fee8a62f4e0de43c20063d250f - Keywords: molecular docking - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Analysis of polypeptide derivati1ves targeting the melanocortin-4 receptor for treating hypoactive sexual desire disorder - Source: Eclética Química (journals) - Date: 2026-09-06T00:00:00Z - Categories: Docking & Screening - Journal: Eclética Química - DOI: 10.26850/1678-4618.eq.v51.2026.e1648 - External ID: 4f328439b828b6492bf22654cfe135ba92dcb940 - Keywords: Molecular docking, pharmacokinetic, binding affinity, Density functional theory, DFT, receptor, Molecular dynamics - Source URL: - Dashboard article: Abstract: Hypoactive sexual desire disorder (HSDD) affects over 13% of individuals globally, significantly impacting emotional well-being and relationships. Current pharmacological treatments for HSDD often have side effects. This study explored 14 peptide-based compounds as potential treatments for HSDD using in silico methods. Molecular docking results identified four compounds with promising potential as alternatives to existing medications. Molecular dynamics (MD) simulations compared the leading compound’s binding affinity to the reference drug’s affinity with the melanocortin-4 receptor (MC4R), showing a greater binding affinity (∆GGBSA = –76.23 kJ/mol) for the leading compound. Compound 8 also exhibited pharmacokinetic properties like the reference drugs. Density functional theory (DFT) analysis revealed that compound 8 had higher reactivity than the reference drug, indicated by a lower HOMO-LUMO value. These findings highlight the potential of compound 8 as a therapeutic agent for HSDD and its promise in developing a new class of peptide-based treatments. ## Computational prediction of flavonoid inhibitors targeting the infectious bronchitis virus spike protein through integrated virtual screening and molecular dynamics - Source: In Silico Pharmacology (journals) - Date: 2026-09-06T00:00:00Z - Journal: In Silico Pharmacology - DOI: 10.1007/s40203-026-00727-z - External ID: e4c4249f3ff96c97b2bb3afd9e8b07169b2ad184 - Keywords: virtual screening, molecular dynamics - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Design, Synthesis, Antimicrobial Evaluation, DFT Analysis, Molecular Docking and Molecular Dynamics Studies of Thiadiazol-Hydrazine Derivatives as Potential Dihydrofolate Reductase Inhibitors. - Source: The protein journal (journals) - Date: 2026-09-06T00:00:00Z - Categories: Docking & Screening - Journal: The protein journal - DOI: 10.1007/s10930-026-10351-7 - External ID: 12173df4f1b2cd7efa257334e7408015865eeb41 - Keywords: Molecular Docking, DFT, Molecular Dynamics - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Design, synthesis, drug-likeness, and molecular dynamics evaluations of monobrominated amathamide G analogs as potential ligands for D2-like dopamine receptors - Source: In Silico Pharmacology (journals) - Date: 2026-09-06T00:00:00Z - Categories: Docking & Screening - Journal: In Silico Pharmacology - DOI: 10.1007/s40203-026-00741-1 - External ID: ed87cd2a0b027232b8e324e008eb710f25b0682e - Keywords: Molecular docking, drug likeness, carcinogenicity, molecular dynamics, pharmacokinetic - Source URL: - Dashboard article: Abstract: Dopamine receptors (DRs) are key modulators of physiological and behavioral responses in the central nervous system (CNS), playing a major role in psychotic disorders. Among these, the D2-like members D2R, D3R and D4R are particularly attractive therapeutic targets for the development of novel compounds derived from both synthetic and natural sources. The marine bryozoan Amathia produces alkaloids known as amathamides (A-H), among which the tribrominated amathamide G (AM-G) stands out as a promising scaffold. Its benzene–aliphatic linker–pentacyclic structure is reminiscent of standard dopamine antagonists such as eticlopride (ETI). In this study, a debrominated (AM-0) and monobrominated derivatives (M4B, M5B and M6B) were successfully synthesized and purified. As an initial preclinical assessment, the drug likeness profiles of these compounds were predicted and analyzed based on physicochemical properties, medicinal chemistry, and pharmacokinetic parameters. The analysis revealed that debromination of AM-G improves both lipid and aqueous solubility, enhances overall drug desirability across four major medicinal chemistry rule sets, and predicts better CNS penetration, bioavailability, half-life, and clearance, while maintaining a relatively low risk of carcinogenicity. Molecular docking and molecular dynamics (MD) simulations suggested a differential affinity amongst amathamide G derivatives towards D2-like receptors. Specifically, AM-G behaved as a high-affinity putative ligand for D3R/D4R, whereas M5B and M6B showed selectivity for D2R. These findings support continued in vitro and in vivo evaluation of these compounds, particularly in models involving expression, structure-based variations, and pathologies associated with D2R-like receptors. ## EGNN ‐Based Generative Models for 3D Molecular Generation - Source: WIREs Computational Molecular Science (journals) - Date: 2026-09-06T00:00:00+00:00 - Categories: Property Prediction, ADMET & Safety, Design de novo - Authors: Siliang Chen, Daihan Wang, Zhaoping Pan, Guobo Li, Gu He - Journal: WIREs Computational Molecular Science - DOI: 10.1002/wcms.70086 - Keywords: Molecular Generation, Generative Models, denoising, graph neural networks, Equivariant, ADMET - Source URL: - Dashboard article: Abstract: Equivariant graph neural networks (EGNNs) are becoming the geometric infrastructure of 3D molecular generation for AI‐aided drug discovery. By enforcing E(n), E(3), or SE(3) equivariance, they separate physical molecular structure from arbitrary coordinate‐frame conventions and ensure that predicted coordinates, denoising directions, and velocity fields transform consistently with molecular geometry. This symmetry‐aware design supports direct modeling of conformations and protein–ligand spatial relationships while remaining compatible with diffusion, flow‐based, autoregressive, and hybrid generative frameworks. Consequently, EGNNs enable more geometrically consistent generation and facilitate controllable design conditioned on binding pockets, pharmacophores, fragments, scaffolds, reference ligands, or molecular properties. Their benefits, however, are bounded by what equivariance encodes. Coordinate‐frame consistency does not itself enforce valid bonds, valence, stereochemistry, topology–geometry agreement, synthesizability, or biological activity. Moreover, practical models must reconcile discrete chemical variables with continuous geometric dynamics, represent long‐range interactions without prohibitive computational cost, and account for protein flexibility, solvent, metal ions, and induced fit. Performance is further constrained by biased docking or property predictors, heterogeneous benchmarks, and limited prospective validation. This review synthesizes how EGNNs function across major generative paradigms and argues that future drug‐oriented systems should combine joint 2D–3D graph generation, physically credible interaction modeling, synthesis and ADMET constraints, multi‐objective optimization, and closed‐loop experimental feedback. EGNNs should therefore be viewed not as a complete chemical solution, but as the symmetry‐aware foundation upon which more testable and biologically relevant molecular design systems can be built. ## Extending Partition Coefficient Predictions from Key Solvent Systems to Biological and Environmental Systems via Linear Solvation Energy Relationship - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-06T00:00:00+00:00 - Categories: Property Prediction, ADMET & Safety - Authors: Zheming Liu, Yan Xu - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c01437 - Source URL: - Dashboard article: Abstract: Accurate prediction of equilibrium partition coefficients of organic compounds in biological and environmental media is critical for evaluating their environmental fate and bioaccumulation potential, as well as for guiding drug design and toxicology studies. Linear solvation energy relationships using Abraham solute descriptors (ASD-LSERs) have achieved remarkable success in characterizing such complex biological and environmental partitioning systems. However, the limited availability of high-quality experimental descriptors, particularly for structurally complex compounds, hinders their practical application and further constrains the accuracy of descriptor predictive models. Here, we propose a linear solvation energy relationship (4SD-LSER) using descriptors derived from logarithmic n-hexadecane–air, n-octanol–air, and water–air partition coefficients, along with the topological McGowan molar volume. To evaluate its performance, 1,849 experimental partition coefficients for 779 neutral compounds across 12 biologically and environmentally relevant systems were compiled. The 4SD-LSER was calibrated and exhibited good descriptive power for these systems. Remarkably, when combined with appropriate fragment-based or machine learning-based descriptor prediction approaches, the 4SD-LSER achieved prediction errors largely within ±0.5 log units for structurally simple compounds and within ±1.0 log unit for more complex compounds (e.g., pesticides, pharmaceuticals, and flame retardants), exhibiting state-of-the-art accuracy, especially for complex compounds. This study demonstrates that models originally developed for well-characterized solvent systems to predict partition coefficients or solvation free energies can be readily extended to biological and environmental systems via LSER. Its performance is poised to improve further with advances in theoretical and machine-learning approaches. ## Extending Partition Coefficient Predictions from Key Solvent Systems to Biological and Environmental Systems via Linear Solvation Energy Relationship - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-06T00:00:00Z - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c01437 - External ID: 12a77c476d8c58539e9863a8a4470d943a0ce411 - Source URL: - Dashboard article: Abstract: Accurate prediction of equilibrium partition coefficients of organic compounds in biological and environmental media is critical for evaluating their environmental fate and bioaccumulation potential, as well as for guiding drug design and toxicology studies. Linear solvation energy relationships using Abraham solute descriptors (ASD-LSERs) have achieved remarkable success in characterizing such complex biological and environmental partitioning systems. However, the limited availability of high-quality experimental descriptors, particularly for structurally complex compounds, hinders their practical application and further constrains the accuracy of descriptor predictive models. Here, we propose a linear solvation energy relationship (4SD-LSER) using descriptors derived from logarithmic n-hexadecane–air, n-octanol–air, and water–air partition coefficients, along with the topological McGowan molar volume. To evaluate its performance, 1,849 experimental partition coefficients for 779 neutral compounds across 12 biologically and environmentally relevant systems were compiled. The 4SD-LSER was calibrated and exhibited good descriptive power for these systems. Remarkably, when combined with appropriate fragment-based or machine learning-based descriptor prediction approaches, the 4SD-LSER achieved prediction errors largely within ±0.5 log units for structurally simple compounds and within ±1.0 log unit for more complex compounds (e.g., pesticides, pharmaceuticals, and flame retardants), exhibiting state-of-the-art accuracy, especially for complex compounds. This study demonstrates that models originally developed for well-characterized solvent systems to predict partition coefficients or solvation free energies can be readily extended to biological and environmental systems via LSER. Its performance is poised to improve further with advances in theoretical and machine-learning approaches. ## Integrated 3D-QSAR, docking and molecular dynamics reveal novel drug-like inhibitors of SARS-CoV-2 main protease - Source: Chemical Papers (journals) - Date: 2026-09-06T00:00:00Z - Categories: Property Prediction - Journal: Chemical Papers - DOI: 10.1007/s11696-026-05518-6 - External ID: 477e8c89bf29e2ebb4de0500583a883649a38ec1 - Keywords: QSAR, molecular dynamics - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Integrated in vitro and in silico evaluations of novel benzanilide derivatives as anticancer agents. - Source: Future medicinal chemistry (journals) - Date: 2026-09-06T00:00:00Z - Categories: Docking & Screening, ADMET & Safety - Journal: Future medicinal chemistry - DOI: 10.1080/17568919.2026.2727415 - External ID: 3a3927639626c663dcd9b0735e5b2c43a33edae4 - Keywords: molecular docking, IC50, enzyme, ADMET, pharmacokinetic - Source URL: - Dashboard article: Abstract: AIMS This study aimed to design, synthesize, and evaluate novel benzanilide derivatives as potential anticancer agents targeting VEGFR-2-mediated tumor angiogenesis. MATERIALS AND METHODS Eight benzanilide derivatives (7a-e, 9a-b, and 11) were synthesized and structurally confirmed using spectroscopic methods. The compounds were evaluated via molecular docking, in silico ADMET profiling, in vitro cytotoxicity against MCF-7, MDA-MB-231, HepG-2, and HCT-116 cancer cell lines, and selectivity assessment using WI-38 and WISH normal cell lines. VEGFR-2 enzyme inhibition was also determined for the most active compound. RESULTS Compound 11 exhibited the highest cytotoxic activity with IC50 values of 6.08, 5.76, 7.94, and 4.64 µM against MCF-7, MDA-MB-231, HepG-2, and HCT-116, respectively, comparable to sorafenib. It showed potent VEGFR-2 inhibition (IC50 = 0.414 µM) superior to sorafenib (0.809 µM). Mechanistic studies indicated activation of apoptosis via upregulation of Bax, Caspase-3, and Caspase-8 and downregulation of Bcl-2. ADMET results revealed favorable pharmacokinetic properties and acceptable safety profiles. CONCLUSIONS Compound 11 represents a promising VEGFR-2 inhibitor with potent anticancer activity and favorable drug-like properties, warranting further optimization and development. ## Machine Learning Guided Discovery of Microbiome Metabolites That Inhibit HDAC - Source: ACS Omega (journals) - Date: 2026-09-06T00:00:00+00:00 - Authors: Hong A. Chung, Emily M. Eshleman, James Carter, Theresa Alenghat, Daniel Reker - Journal: ACS Omega - DOI: 10.1021/acsomega.5c12875 - Source URL: - Dashboard article: Abstract: Histone deacetylase (HDAC) is a family of key epigenetic regulator implicated in inflammation, metabolism, and cancer. Microbiome metabolites can modulate host histone acetylation, yet systematic identification of metabolites that target HDAC remains limited. Here we present a computationally driven discovery pipeline that combines training data composition optimization with experimental validation to prioritize microbiome-derived HDAC inhibitors. Starting from a large, public HDAC3 screening data set (314,129 compounds; 485 actives), we developed an automated iterative sampling strategy that balances active and inactive compounds while enriching the inactive class for metabolite-like chemistry. Models trained on the balanced metabolite-enriched subsets achieved substantially higher sensitivity and balanced accuracy than models trained on the full data set. Consensus predictions from multiple optimized runs were applied to a curated microbiome metabolite database to prioritize candidates for testing. Two top candidates, 5-(hydroxymethyl)furoic acid (HMFA) and d-glucuronolactone (DGL), were evaluated using an HDAC3-specific biochemical assay, followed by broader HDAC activity assessment. Both metabolites exhibited mild inhibitory activity overall. Docking simulations suggested that HMFA may interact with the HDAC3 catalytic site. Together, we show that targeted training set composition can improve machine learning-assisted discovery of microbiome-derived small molecule inhibitors and identified HMFA as a microbiome-associated metabolite with measurable in vitro HDAC inhibitory activity. ## Multi-technique spectroscopic and computational modeling approaches for elucidation of the binding mechanism between two disperse azo dye derivatives and DNA. - Source: Nucleosides, nucleotides & nucleic acids (journals) - Date: 2026-09-06T00:00:00Z - Categories: Docking & Screening - Journal: Nucleosides, nucleotides & nucleic acids - DOI: 10.1080/15257770.2026.2727418 - External ID: 13d19943b740c7e20ed171d863ab1131a2f655fd - Keywords: molecular docking - Source URL: - Dashboard article: Abstract: Using multi-spectroscopy and molecular docking techniques, the binding mechanism between two azo dyes and the calf thymus (Ct-DNA) was investigated under physiological settings. After the addition of Ct-DNA, each azo dye exhibited a hypochromic effect and slightly increased the wavelength of the maximum absorption. This suggested that the probes and Ct-DNA interacted via a groove binding mode, which was corroborated by the molecular docking data. The thermodynamic parameters, ΔG°, ΔH°, and ΔS°, were determined by calculating the binding constants from the maximum absorption spectra of both azo dyes at different Ct-DNA concentrations at different temperatures. Moreover, fluorescence resonance energy transfer indicates that the interval between the donor (EB-Ct-DNA) and acceptor (azo dye) is suitable for energy transfer. Hydrogen bonds and π electrons on the azo dye's benzene ring are essential for binding the dye to Ct-DNA, according to molecular modeling research. These probes bind to the minor groove of Ct-DNA, and the molecular docking results are in good agreement with the spectroscopic findings. ## Numerical Prediction of Physicochemical Properties of Drug Structures via Some Graph Parameters - Source: Symmetry (journals) - Date: 2026-09-06T00:00:00Z - Categories: Property Prediction - Journal: Symmetry - DOI: 10.3390/sym18091492 - External ID: 7da50e67dee15797a7889b0f95da121a4cb4ed02 - Keywords: QSPR - Source URL: - Dashboard article: Abstract: Graph theory has become a fundamental mathematical framework with broad applications across various scientific fields, such as biology, network security, computer science, and chemistry. A specialized branch known as chemical graph theory employs graph-based mathematical principles to model and analyze molecular architectures. Typically, molecular graphs are generated from 2D representations of chemical structures, where physicochemical properties play a decisive role in interpreting a molecule’s physical and chemical behavior. Structural symmetry within these molecular graphs significantly aids in deriving fundamental graph parameters. Recently, a novel theoretical method has been introduced to perform a comparative analysis of seven significant domination parameters alongside the p-electronic energy of benzenoid hydrocarbons. In the field of graph theory, the independence, covering, and domination numbers stand as fundamental parameters. Inspired by this methodology and related research, this study presents an exploratory investigation into the physicochemical properties of 10 selected drug and bioactive molecules from diverse therapeutic classes to evaluate bulk size-dependent properties using these fundamental graph parameters. Various regression models were evaluated alongside leave one out cross validation (LOOCV) diagnostics for six specific properties of these compounds: boiling point (BP), enthalpy of vaporization (EV), flash point (FP), molar refractivity (MR), polarizability (P), and molar volume (MV). The results highlight the potential of these fundamental graph parameters in QSPR modeling, establishing an exploratory proof of concept that warrants future support from more comprehensive independent datasets. ## Ranking-Based Surrogate Modeling for Bayesian Optimization under Small-Data Conditions - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-06T00:00:00+00:00 - Categories: Reaction Informatics - Authors: Yuya Endo, Hiromasa Kaneko - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c02115 - Source URL: - Dashboard article: Abstract: Bayesian optimization (BO) is widely used for reaction optimization, but under small-data conditions, regression of absolute objective values may not align with the practical goal of prioritizing promising experiments. Here, we compared ranking-based BO (RankBO) combined with Thompson sampling (Rank-TS) with regression-based BO using two reaction benchmarks. Rank-TS showed a clear advantage on Direct Pd-catalyzed arylation in optimization performance, global ranking quality, and recovery of high-yielding conditions. For Suzuki–Miyaura coupling, which comprised 12 substrate combinations, its improvement was small in the pooled analysis. However, in the equal-weight macro analysis across the 12 substrate-defined search spaces, Rank-TS showed higher ranking quality and faster high-yield-condition recovery. These results indicate that Rank-TS is particularly useful for prioritizing reaction conditions within a fixed-substrate combination under small-data conditions, and that its relative performance depends on how the chemical search space and ranking task are defined. ## STRUCTURAL CHARACTERIZATION AND COMPARATIVE IN SILICO SCREENING OF POTENTIAL LIGANDS AGAINST ANTIMICROBIAL RESISTANCE PROTEINS IDENTIFIED FROM THE RIVER VARUNA - Source: Genetics and Molecular Research (journals) - Date: 2026-09-06T00:00:00Z - Categories: Docking & Screening - Journal: Genetics and Molecular Research - DOI: 10.4238/w5h4b069 - External ID: 9f7f99616e15f64b0629d402fb625b035993ee11 - Keywords: molecular docking, IN SILICO SCREENING, computational screening - Source URL: - Dashboard article: Abstract: Antimicrobial resistance (AMR) in aquatic environments represent an important environmental dimension of the global AMR load, specifically in the polluted river systems receiving diverse anthropogenic inputs. The present study investigated selected abundant AMR proteins identified at four sites (VR1-VR4) of the river Varuna using a whole-genome shotgun metagenomic approach. Further, a computational screening-ligand-associated study was performed. AMR protein sequences were subjected to BLASTp analysis for sequence-level identification, followed by conserved domain analysis to confirm their functional properties. The proteins identified were Sul1, Sul4, blaOXA, and blaVEB-9 from sites VR1, VR2, VR3, and VR4. Conserved domain analysis identified dihydropteroate synthase domain in Sul1 and Sul4, and a class D ß-lactamase domain in blaOXA and a class A ß-lactamase related serine hydrolase domain in blaVEB-9. Experimentally determined 3D protein structures were retrieved from the Protein Data Bank termed as 7S2I, 1TWS, 1M6K, and 6NVT. Comparative cavity-guided molecular docking was performed using CB-Dock2 with sulfamethoxazole (FDA-approved drug) and thymol (natural compound). Thymol showed predicted docking scores ranging from -6.3 to -8.1kcal/mol. The most favorable predicted interaction for both ligands was observed with blaVEB-9. This finding provides preliminary structural insights into ligand-binding patterns of environment-derived AMR (Antimicrobial resistance) proteins and identifies protein-ligand combinations to analyze further biochemical and microbiological validation. ## Structure-based virtual screening, multi-score docking, and molecular dynamics simulation of novel small molecules targeting the epidermal growth factor receptor for potential management of oral squamous cell carcinoma - Source: In Silico Pharmacology (journals) - Date: 2026-09-06T00:00:00Z - Journal: In Silico Pharmacology - DOI: 10.1007/s40203-026-00722-4 - External ID: 51d8f1ef90de595744f032b07fc1ef1d25181d66 - Keywords: virtual screening, receptor, molecular dynamics - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Agents in Drug Discovery – A few Thoughts - Source: DrugDiscovery.NET (feeds) - Date: 2026-09-05T11:55:14+00:00 - Categories: Blog - Source URL: - Dashboard article: - Abstract: not stored for this record. ## A dynamic fusion diffusion model for molecular generation. - Source: Molecular diversity (journals) - Date: 2026-09-05T00:00:00Z - Categories: Design de novo - Journal: Molecular diversity - DOI: 10.1007/s11030-026-11718-9 - External ID: 91ff881d12779d9f9059989f58cce8d9ca25568b - Keywords: molecular generation, diffusion model - Source URL: - Dashboard article: - Abstract: not stored for this record. ## A fuzzy graph-based Kulli–Basava descriptor for isomer discrimination and physicochemical property prediction - Source: Scientific Reports (journals) - Date: 2026-09-05T00:00:00+00:00 - Categories: Property Prediction - Authors: Iqra Yaqoot, Zeeshan Saleem Mufti, Abdulrahman A. Almehizia, Ahmad A. Assiri, Gamachu Adugna Ganati - Journal: Scientific Reports - DOI: 10.1038/s41598-026-68990-w - Keywords: QSPR, property prediction, molecular descriptor - Source URL: - Dashboard article: Abstract: In this paper, a generalized fuzzy Kulli-Basava index ( $$FKB^\{\*\}\_1$$ ) of fuzzy graphs is introduced, which is an extension of the classical index, by introducing a fuzzy vertex degree, fuzzy edge membership value, and fuzzy sum of the degree of incident vertices. Unlike existing fuzzy topological descriptors such as the fuzzy Randić index ( $$\\textrm\{FR\}$$ ) and fuzzy first Zagreb index ( $$\\textrm\{FZ\}\_1$$ ), which rely solely on endpoint degrees, $$FKB^\{\*\}\_1$$ incorporates $$I\_e(v)$$ , a vertex-weight that aggregates fuzzy degrees of neighbours and reaches two bonds into the molecular environment, thereby capturing richer structural information. When considering binary membership, the index reduces to $$KB\_1$$ and is found to perfectly discriminate between isomers in the case of the three alkane families (hexane, heptane, and octane), outperforming all classical indices, with the exception of assigning distinct values to most isomers of octane. QSPR analysis demonstrates exceptional correlation with acentric factor and strong performance for entropy. The computational algorithm ensures efficiency for molecular graphs, and the index is most effective within fixed carbon chain lengths for properties related to the molecular branching. $$FKB^\{\*\}\_1$$ is a powerful molecular descriptor but is not a complete graph invariant. ## A novel framework for the discovery of MAPK-activated protein kinase 2 (MAPKAPK2) inhibitors using a multi-feature deep learning ensemble - Source: Journal of Computer-Aided Molecular Design (journals) - Date: 2026-09-05T00:00:00+00:00 - Authors: Hayden Chen, Yi-Wen Wu, Tony Eight Lin, Jun-Hong Chen, Yu-Cheng Chan, Chun-Lin Yang, Shih-Chung Yen, Wei-Chun HuangFu, Shiow-Lin Pan, Kai-Cheng Hsu - Journal: Journal of Computer-Aided Molecular Design - DOI: 10.1007/s10822-026-00935-x - Keywords: kinase - Source URL: - Dashboard article: - Abstract: not stored for this record. ## AI-Driven Rational Design of Nanotherapeutics for Pancreatic Ductal Adenocarcinoma. - Source: Medicinal research reviews (journals) - Date: 2026-09-05T00:00:00Z - Journal: Medicinal research reviews - DOI: 10.1002/med.70105 - External ID: 0e756d76d6f2a9c8c7eb5e789d49acf701141ee8 - Source URL: - Dashboard article: Abstract: Pancreatic ductal adenocarcinoma (PDAC) remains a formidable malignancy characterized by late diagnosis, high recurrence rates, and pronounced chemoresistance. While nanoparticle-based drug delivery systems (NDDS) offer theoretical advantages over conventional therapies, their clinical translation in PDAC has been severely limited. The dense desmoplastic stroma, elevated interstitial fluid pressure, and hypovascularity effectively neutralize the enhanced permeability and retention (EPR) effect, restricting passive nanoparticle penetration. Consequently, the therapeutic paradigm is shifting from indiscriminate stromal depletion-which paradoxically accelerates metastasis-to stromal normalization and immune microenvironment reprogramming. This review comprehensively evaluates advanced NDDS platforms, highlighting transcytosis-triggering active targeting, stimuli-responsive carriers, and multimodal theranostic systems designed to bypass physical barriers and convert immunologically "cold" tumors to "hot." Furthermore, we analyze the expanding role of artificial intelligence (AI) and machine learning in predicting synergistic drug combinations and optimizing nanoparticle physicochemical properties. Crucially, we critique current epistemic limitations in AI-driven nanomedicine, particularly data sparsity and the reliance on 2D preclinical models, emphasizing the necessity of 3D patient-derived organoid validation. By integrating rational NDDS design with AI and biomarker-driven strategies, this review maps the essential pathways toward personalized nanotherapeutics for PDAC. ## Characteristic flavors of different processed Yunnan specialty coffee after hot brewing: From sensory attributes to molecular recognition mechanism. - Source: Food chemistry (journals) - Date: 2026-09-05T00:00:00Z - Categories: Docking & Screening - Journal: Food chemistry - DOI: 10.1016/j.foodchem.2026.151050 - External ID: 0f2b37313462e90b80a3f3b9ce8b1a7f8f745c57 - Keywords: Molecular docking - Source URL: - Dashboard article: Abstract: This study systematically investigated the characteristic flavors of hot-brewed Yunnan specialty coffees subjected to different primary processing methods (washed, honey-processed, and sun-dried). An integrated multi-scale approach was employed, combining sensory evaluation, physicochemical analysis, chromatography-mass spectrometry (LC-MS/MS and HS-SPME-GC-MS), and computational chemistry. Twenty-three aroma-active compounds (OAV > 1) were identified as major contributors. Aroma recombination and omission experiments confirmed their sensory significance, establishing a quantitative link between chemical composition and sensory attributes. Molecular docking and dynamics simulations revealed that key aroma molecules (e.g., isopulegol, nonanal, ethyl decanoate) primarily interact with olfactory receptors OR1A1 and OR8D1 via van der Waals forces and hydrophobic interactions, elucidating their aroma presentation mechanisms. This research provides a comprehensive data chain and methodological framework for understanding and optimizing the flavor quality of Yunnan specialty coffee through targeted processing and brewing techniques. ## Characterization of Netupitant Degradation Products Employing LC-QTOF-MS and Computational Tools - Source: Asian Journal of Chemistry (journals) - Date: 2026-09-05T00:00:00Z - Categories: Docking & Screening - Journal: Asian Journal of Chemistry - DOI: 10.14233/ajchem.2026.36339 - External ID: 563966f91a75547b27a2d1ce6d3b0f4bd98d55c6 - Keywords: molecular docking, binding affinity, receptor, pharmacokinetic - Source URL: - Dashboard article: Abstract: The present study evaluates the stability of netupitant under various environmental stress conditions and identifies the degradation products formed during the degradation process. In addition, the study predicts the pharmacokinetic and toxicological properties of these degradation products using advanced analytical and computational approaches. Isocratic liquid chromatography was used to separate stress samples, high-resolution quadrupole time-of-flight mass spectrometry was used to elucidate the structure of these samples. Computational platforms were used in making predictions of toxicity and molecular docking simulations were performed to assess the relative binding affinities of the parent drug and its degradants to the human neurokinin-1 receptor. Netupitant was susceptible to oxidative and hydrolytic stress, mainly through oxidative cleavage of the piperazine ring. Molecular docking showed that netupitant has the best neurokinin-1 receptor binding affinity (-12.1 kcal/mol) and structural loss in the degradants slightly alters the binding affinity towards the receptor (between -11.3 and -10.2 kcal/mol). Shifting the degradation products through software screens comprising the pkCSM webserver, ToxTree and OSIRIS property explorer yielded predictions for various pharmacokinetic and toxicity-related parameters. The computational predictions highlighted high intestinal absorption and hepatotoxicity across all compounds, with specific mutagenic structural alerts identified in one degradation product and a tumorigenic alert in another. ## Chemometric insights into Lactiplantibacillus plantarum effects on onion (Allium cepa L.) metabolism and antidiabetic activity under cadmium stress - Source: Scientific Reports (journals) - Date: 2026-09-05T00:00:00+00:00 - Authors: Salwa M. Abdel Rahman, Nancy M. El Halfawy, Rahma S. R. Mahrous - Journal: Scientific Reports - DOI: 10.1038/s41598-026-67251-0 - Keywords: Chemometric, enzyme, OPLS - Source URL: - Dashboard article: Abstract: Cadmium (Cd) is a toxic heavy metal that causes severe physiological damage in plants, inhibiting growth and ultimately reducing crop yield. Lactic acid bacteria regulate Cd availability through bioaccumulation and biosorption. This study evaluated the Cd tolerance of Lactiplantibacillus plantarum 10CH by determining its survival capacity under Cd stress and its potential to mitigate Cd-induced stress in onion ( Allium cepa L.). The bacterial strain tolerated Cd concentrations up to 100 µM, and whole-genome sequencing identified genes involved in Cd biosorption, accumulation, and efflux. Exposure of onion to increasing CdCl 2 concentrations significantly reduced root and shoot biomass. Inoculation with Lb. plantarum 10CH alleviated Cd stress at 100 µM, enhancing root and shoot biomass, reducing Cd accumulation, lowering oxidative damage markers, and stimulating antioxidant enzyme activities. Metabolic profiling revealed that Cd stress significantly reduced primary metabolites and amino acids, particularly at 100 µM, while bacterial inoculation restored key amino acids and peptides, including arginine, tyrosine, and glutamic acid. Chemometric analysis using unsupervised (PCA) and supervised (OPLS-DA) models revealed clear metabolite variation among untreated, Cd-stressed, and bacterial inoculated Cd-stressed onion leaves. Furthermore, leaf extracts exhibited α-glucosidase inhibitory activity, with the highest activity in control plants (IC 50 = 425.2 ± 0.5 µg/mL). Cd-stressed plants showed moderate antidiabetic activity, which was significantly reduced by bacterial inoculation. Overall, these findings demonstrate that Lb. plantarum 10CH can survive under Cd stress and alleviates Cd-induced stress in onion, highlighting its potential as a bioinoculant to mitigate heavy metal stress. ## Computational Exploration of the Structural, Electronic and Nonlinear Optical Properties of Substituted Piperidine Scaffold: A DFT and Molecular Docking Approach - Source: Asian Journal of Chemistry (journals) - Date: 2026-09-05T00:00:00Z - Categories: Docking & Screening - Journal: Asian Journal of Chemistry - DOI: 10.14233/ajchem.2026.36470 - External ID: 8ba747145e29697666bcc7936225ac48100996e8 - Keywords: Molecular Docking, DFT, density functional theory, B3LYP - Source URL: - Dashboard article: Abstract: An investigation into the structural, electronic and reactive properties of a substituted piperidine scaffold were conducted using density functional theory (DFT) at the B3LYP/6-31 G(d) level of theory. Molecular geometry optimization established a stable equilibrium structure. Frontier molecular orbital (FMO) analysis suggested a HOMO-LUMO energy gap of 5.6665 eV, consistent with high kinetic stability. Molecular electrostatic potential (MEP) mapping and Mulliken charge analysis identified the oxygen atoms as the principal nucleophilic centers, while the hydroxyl hydrogen bears a pronounced electrophilic character. Natural bond orbital (NBO) and non-covalent interaction (NCI) analyses quantified pronounced hyperconjugative stabilization and a complex network of hydrogen bonding interactions, particularly across the hydrazone linkage, which exhibited a Mayer bond order of 1.6852. The first-order hyper-polarizability was calculated as 1.044 × 10–30 esu, approximately 2.8-fold higher than that of urea, supporting the potential of the molecule for nonlinear optical (NLO) applications. Thermodynamic parameters exhibited strong temperature dependence, while STM and ALIE surface analyses identified the nitrogen-rich core as an electronically active region. Molecular docking against the 1BP1 protein yielded a binding energy of -6.4 kcal/mol, with hydrogen-bonding interactions involving VAL A:433 and PRO A:430. These computational findings support the relevance of the molecular scaffold for further investigation in medicinal chemistry and optoelectronic applications. ## Controllable molecular generation with fine-tuned flow-matching model - Source: Communications Chemistry (journals) - Date: 2026-09-05T00:00:00+00:00 - Categories: Design de novo - Authors: Kunyu Wang, Jon Paul Janet, Alessandro Tibo - Journal: Communications Chemistry - DOI: 10.1038/s42004-026-02188-z - Keywords: molecular generation, conditional generation, reinforcement learning, generative models - Source URL: - Dashboard article: Abstract: Three-dimensional molecular generative models have emerged that produce de novo molecules both unconditionally and conditionally, e.g., within protein pockets. However, steering those models in a specific region of the chemical space that satisfies a set of desired properties remains challenging. In this study, we introduce a flexible reinforcement learning method for flow-matching based generative models, allowing the velocity field to be refined according to a user-defined reward function. In contrast to a pure conditional generation setup, where the set of conditions must be decided a priori, this framework allows fine-tuning of any unconditional or conditional model, reflecting a more realistic scenario where the target properties to be optimized often vary and are typically case-specific. This also enables joint optimization of continuous and discrete features in flow-matching models for the first time. Through extensive experiments across diverse optimization scenarios, we demonstrate that models trained with this strategy ( agents ) consistently outperform baseline approaches ( priors ) when evaluated against the target design criteria. ## Darwinian Nanomedicine: Artificial Intelligence-Driven Evolutionary Drug Delivery for Adaptive and Precision Therapeutics. - Source: Drug development and industrial pharmacy (journals) - Date: 2026-09-05T00:00:00Z - Journal: Drug development and industrial pharmacy - DOI: 10.1080/03639045.2026.2729377 - External ID: 4775f1197488327f8055fcbf82837c34ddfe552f - Source URL: - Dashboard article: Abstract: OBJECTIVE This review presents Darwinian nanomedicine as an artificial intelligence (AI)-enabled drug delivery strategy that applies Darwinian evolutionary principles as a computational and engineering analogy. Rather than implying biological evolution of nanoparticles, the framework employs iterative cycles of variation, selection, adaptation, and computational inheritance to optimize nanoparticle design through experimental feedback and AI-driven optimization. SIGNIFICANCE OF REVIEW Conventional nanocarrier-based drug delivery systems have improved bioavailability, targeting, and therapeutic efficacy but remain static in design and limited in their ability to accommodate disease heterogeneity and variability. Darwinian nanomedicine represents a framework that integrates artificial intelligence with evolutionary optimization principles to iteratively refine nanoparticle formulations based on experimental performance rather than biological evolution. KEY FINDINGS Recent studies demonstrate that diverse nanoparticle libraries can be iteratively screened and computationally optimized according to performance. Advances in artificial intelligence, machine learning, high-throughput screening, Bayesian optimization, and evolutionary algorithms have accelerated the identification of nanoparticle formulations with improved targeting efficiency, drug release, stability, and performance through closed-loop design-test-learn workflows. CONCLUSIONS Darwinian nanomedicine represents a paradigm shift from static nanoparticle design toward AI-guided adaptive optimization of drug delivery systems. Importantly, the evolutionary processes described in the framework are computationally inspired rather than biological, relying on iterative engineering and experimental optimization instead of nanoparticle self-replication or genetic inheritance. Although significant challenges remain regarding scalability, reproducibility, data standardization, manufacturing, and regulatory approval, continued advances in nanotechnology and artificial intelligence may facilitate the future translation of this framework into precision medicine applications. ## Design, Synthesis and Anti-Breast Cancer Activity of New Benzimidazole-1,2,4-thiadiazoles-benzamide Hybrids - Source: Asian Journal of Chemistry (journals) - Date: 2026-09-05T00:00:00Z - Categories: Docking & Screening - Journal: Asian Journal of Chemistry - DOI: 10.14233/ajchem.2026.36472 - External ID: abd4977d968a1361c7b5339a6f818f15f387ab7e - Keywords: Molecular docking, kinase, receptor - Source URL: - Dashboard article: Abstract: A new series of benzimidazole-1,2,4-thiadiazole-benzamide hybrids (7a-l) was synthesized and characterized. The synthesized hybrids were evaluated for anticancer activity against two breast cancer cell lines, MCF-7 and MDA-MB-231, using the MTT assay, with Erlotinib as the reference drug. The results revealed that hybrid 7i exhibited more potent activity against both tested cancer cell lines than the standard drug Erlotinib. The other three hybrids, 7j, 7k and 7l, displayed activity values close to that of the standard drug. Based on these findings, hybrids 7i, 7j, 7k and 7l were further evaluated for tyrosine kinase EGFR inhibitory activity using Erlotinib as the reference drug. Hybrids 7i and 7j exhibited higher EGFR inhibitory activity than erlotinib. Molecular docking studies of the potent hybrids with the EGFR receptor supported the biological results, with hybrids 7i and 7l exhibiting strong interactions with the EGFR protein and superior binding affinities compared with erlotinib. The combined anticancer, EGFR inhibitory and docking results identify hybrid 7i as the most promising hybrid, while hybrids 7j and 7l represent potential EGFR-targeting anticancer candidates. ## Design, Synthesis and in vitro Evaluation of Phthalimide-Anhydride Derivatives as MMP-9 Suppressors in Breast Cancer Therapy - Source: Asian Journal of Chemistry (journals) - Date: 2026-09-05T00:00:00Z - Categories: Docking & Screening - Journal: Asian Journal of Chemistry - DOI: 10.14233/ajchem.2026.36182 - External ID: 0e23157961508bbb03fe5fd2b8e34ebf21479a2d - Keywords: Molecular docking, Lipinski, Glide score, IC50, AMES, hERG, Molecular dynamics - Source URL: - Dashboard article: Abstract: A novel series of substituted acetic 2-(1,3-dioxo-1,3-dihydro-2H-isoindol-2-yl)anhydride derivatives (5a-e) was designed and evaluated as potential matrix metalloproteinase-9 (MMP-9) suppressors for breast cancer therapy. Molecular docking using Schrödinger Maestro 2021 (PDB ID: 6ESM) revealed effective chelation of the catalytic Zn2+ ion by all derivatives, with the highest Glide score of -5.900 kcal/mol for compound 5a and a promising -4.632 kcal/mol for lead candidate 5c \[acetic 2-(1,3-dioxo-1,3-dihydro-2H-isoindol-2-yl)-3-phenylpropanoic anhydride\]. Molecular dynamics simulations confirmed the compound 5c–MMP-9 complex stability (RMSD: 1.5-2.5 Å over 100 ns). SwissADME predictions indicated drug-like properties, including high gastrointestinal absorption, compliance with Lipinski’s rule of five and no AMES mutagenicity or hERG inhibition. Compounds were synthesised via a two-step protocol involving phthalic anhydride condensation with amino acids followed by acetyl chloride-mediated cyclodehydration, affording yields of 46-65% (compound 5c: 65%). In vitro, compound 5c exhibited moderate cytotoxicity against MCF-7 breast cancer cells (IC50 = 27.89 µg/mL; 82.7 µM), approaching that of 5-fluorouracil (IC50 = 8.6 µg/mL; 66.1 µM) in the same assay system. Flow cytometry showed that compound 5c induced early apoptosis (30.6% vs. 9.3% controls) and little necrosis. ELISA tests indicated that there was a concentration-dependent reduction of MMP-9 in TPA-stimulated MCF-7 cells (64.36% at a concentration of 55.89 µg/mL, 165.7 µM). Acute oral toxicity in rats gave an LD50 cut-off of 5-50 mg/kg (GHS Category 2), limiting the therapeutic window and requiring further optimisation before preclinical development. The phthalimide-anhydride scaffold offers a promising Zn2+-chelating motif with potential selectivity over hydroxamates. Compound 5c merits further lead optimisation, with in vivo efficacy, pharmacokinetics and selectivity against other MMPs requiring evaluation. ## Development of Netrin-1-targeted peptide probes derived from the Netrin-1/DCC complex for noninvasive imaging of tumors. - Source: Bioorganic chemistry (journals) - Date: 2026-09-05T00:00:00Z - Journal: Bioorganic chemistry - DOI: 10.1016/j.bioorg.2026.110479 - External ID: 641814de2784a1475a1f56b6156768610fe31f3a - Keywords: molecular docking, binding affinity - Source URL: - Dashboard article: Abstract: Netrin-1 is a secreted glycoprotein that is overexpressed in non-small cell lung cancer (NSCLC) and breast cancer, making it an emerging biomarker for tumor diagnosis and targeted therapy. However, noninvasive imaging tools for quantitative assessment of Netrin-1 expression remain limited. In this study, leveraging the crystal structure of the Netrin-1/DCC complex, we rationally designed two Netrin-1-targeted peptides YP8 and LE7, through systematic in silico molecular docking, alanine scanning, and virtual amino acid mutation. Their corresponding 68Ga-labeled PET tracers \[68Ga\]Ga-NOTA-YP8 and \[68Ga\]Ga-NOTA-LE7, were subsequently developed. Comparative evaluation revealed that \[68Ga\]Ga-NOTA-LE7 exhibited superior Netrin-1 binding affinity, enhanced in vivo stability, and significantly improved tumor-to-background ratio. PET imaging demonstrated specific and robust uptake of \[68Ga\]Ga-NOTA-LE7 in Netrin-1-positive A549 and 4T1 tumors, with minimal accumulation in Netrin-1-low MDA-MB-231 tumors. The NIRF probe ICG-LE7 further confirmed specific tumor targeting and enabled fluorescence-guided surgical resection. Collectively, these findings establish \[68Ga\]Ga-NOTA-LE7 and ICG-LE7 as promising noninvasive diagnostic tools for Netrin-1 expression imaging. ## Exploration of novel anti-TB agents targeting the serine/threonine kinase enzyme (PknE) in Mycobacterium tuberculosis: an in silico study - Source: Scientific Reports (journals) - Date: 2026-09-05T00:00:00+00:00 - Categories: Docking & Screening, ADMET & Safety, Targets & Structures, Free Energy & MD - Authors: Abdallah Elssir Ahmed, Reem M. A. Ebrahim, Nooh Mohamed Hajhamed, Lamis Yahia Mohamed Elkheir, Sarmad Marah, Hiba Hassan Sulieman Omer, Tevfik Ozen, Nouh Saad Mohamed, Mohammed H. Abdelraheem, Ayman Azhary, Faisal Hammad Mekky Koua, Mohammed Asaad, Asim Osman. Abdoun, Waleed Abdelateif Hussein, Seedahmed A. Mohammed, Amar Mohamed Ismail - Journal: Scientific Reports - DOI: 10.1038/s41598-026-67951-7 - Keywords: molecular docking, virtual screening, kinase, enzyme, receptor, binding free energy, free energy, Gibbs free energy, molecular dynamics, MD simulations - Source URL: - Dashboard article: Abstract: Mycobacterium tuberculosis (MTB) remains a deadly infectious agent and a global health challenge, particularly due to the emergence of multidrug-resistant strains. Mycobacterial serine/threonine protein kinase E (PknE) is vital for mycobacterial survival under nitric oxide stress by altering Toll-like receptor expression, suppressing apoptosis, increasing inflammation, and modulating costimulatory molecules. This study employed a structure-based in silico workflow comprising virtual screening of 2,202 FDA-approved compounds, molecular docking, 300 ns molecular dynamics (MD) simulation molecular mechanics/Poisson-Boltzmann surface area (MM/PBSA) binding free energy calculations, principal component analysis (PCA), free energy landscape (FEL) analysis, and in silico toxicity profiling, to identify candidate PknE inhibitors. Molecular docking identified netilmicin, dirithromycin, acarbose, and ertapenem as top candidates, with docking scores of − 13.13 to − 15.86 kcal/mol. MD simulations confirmed complex stability with RMSD values near 1 nm, while MM/PBSA analysis revealed favourable binding free energies: netilmicin (− 1.47 ± 0.18), ertapenem (− 11.70 ± 1.22), dirithromycin (− 17.80 ± 0.13), and acarbose (− 18.90 ± 0.06 kcal/mol). PCA-based FEL analysis confirmed thermodynamic stability across all complexes, with DTM–PknE exhibiting the lowest minimum Gibbs free energy (15.7 kJ/mol). In silico toxicity profiling demonstrated acceptable safety profiles for all leads. These findings provide a strong computational basis for repurposing these agents as adjunct anti-TB therapies, pending experimental validation. ## Exploration of Novel Thiazole Scaffolds with Potential Therapeutic Applications - Source: Asian Journal of Chemistry (journals) - Date: 2026-09-05T00:00:00Z - Categories: Docking & Screening - Journal: Asian Journal of Chemistry - DOI: 10.14233/ajchem.2026.36430 - External ID: 3c58a85a449478d0fdf06160823d5b086367b631 - Keywords: molecular docking, AutoDock - Source URL: - Dashboard article: Abstract: This study focuses on the synthesis and evaluation of the antimicrobial, anti-inflammatory and antipyretic activities of novel thiazole-based Schiff base derivatives (S1-S6). The compounds were synthesized through condensation reactions between 2-aminothiazole and several aromatic aldehydes. Structural characterization of the synthesized derivatives was carried out using FTIR, 13C NMR, 1H NMR, mass spectrometry techniques and elemental analysis. Initially, thiazole-based Schiff base derivatives therapeutic activities were assessed from AutoDock score after the successful binding with Staphylococcus aureus DNA gyrase B and cyclooxygenase 2. The synthesized derivatives were also evaluated for antimicrobial, anti-inflammatory and antipyretic activities using in vitro and in vivo methods. Compounds S6 and S1 showed the highest activity, possibly due to the thiazole nucleus linked to furan and substituted aromatic rings. The experimental findings were consistent with the in silico molecular docking results. ## GraphPert: A graph autoencoder framework for network-based drug repurposing using transcriptional signatures-A case study on melanoma. - Source: Computers in biology and medicine (journals) - Date: 2026-09-05T00:00:00Z - Journal: Computers in biology and medicine - DOI: 10.1016/j.compbiomed.2026.111912 - External ID: fb2fa2e9728c1a303da212e278a369221d477739 - Keywords: virtual screening, autoencoder - Source URL: - Dashboard article: Abstract: MOTIVATION Network-based drug repurposing leverages the principle that drug effects propagate through protein-protein interaction (PPI) networks similarly to disease perturbations. Graph-based embedding methods exploit this principle by learning low-dimensional representations that capture the network neighborhood of disease and drug target proteins, enabling similarity-based prioritization of therapeutic candidates. However, existing approaches that learn fixed embeddings from a static graph cannot accommodate perturbations to the graph structure without costly recomputation or retraining, making them impractical for large-scale screening where each drug generates a distinct transcriptional perturbation. OBJECTIVE We present GraphPert, a Graph Autoencoder (GAE) framework that propagates transcriptional signatures through the human PPI to enable efficient network-based virtual screening. METHODS Transcriptional signatures from the Connectivity Map L1000 platform are encoded as symmetric edge-weight modifications in the PPI, scaling each interaction by the expression changes of both participating proteins. A GAE pre-trained on the human PPI (N=12,458 proteins) produces latent displacements for each perturbation, and compounds are ranked by cosine similarity to a therapeutically relevant reference-the BRAF V600E knockout (KO) in A375 melanoma cells. RESULTS In retrospective screening against a decoy library of known MAPK cascade inhibitors seeded among compounds without known MAPK pathway activity (1:9 ratio), GraphPert achieves AUC >0.80. Notably, while raw signature-based screening recovers predominantly compounds with targets proximal to the MAPK pathway, GraphPert additionally identifies compounds with more distal targets-including HDAC, proteasome, and PARP inhibitors-that converge on the same downstream functional modules suppressed by BRAF KO. CONCLUSION GraphPert demonstrates that propagating transcriptional signatures through a pre-trained GAE extends the reach of virtual screening beyond direct transcriptional similarity, enabling the discovery of mechanistically diverse compounds that recapitulate the network-level effects of a therapeutic reference perturbation. ## Impact of cigarette toxicants on sarcopenic obesity, Insights from Network Toxicology, Molecular Docking and Dynamics - Source: Tobacco Induced Diseases (journals) - Date: 2026-09-05T00:00:00Z - Categories: Docking & Screening - Journal: Tobacco Induced Diseases - DOI: 10.18332/tid/226887 - External ID: a098240e9dae5435e6e2bf3c70db47ad210360f6 - Keywords: Molecular Docking - Source URL: - Dashboard article: - Abstract: not stored for this record. ## IN SILICO TOXICITY PROFILING OF BENZOPHENONE-3 WITH FOCUS ON DEVELOPMENTAL AND METABOLIC PATHWAYS - Source: Ankara Universitesi Eczacilik Fakultesi Dergisi (journals) - Date: 2026-09-05T00:00:00Z - Categories: Docking & Screening - Journal: Ankara Universitesi Eczacilik Fakultesi Dergisi - DOI: 10.33483/jfpau.1747733 - External ID: 88b420d8d47b35e49e1ca89d490d113209b1d127 - Keywords: molecular docking, binding affinity, cytochrome P450 - Source URL: - Dashboard article: Abstract: Objective: This study aims to evaluate the developmental toxicity potential of Benzophenone-3 (BP-3), a widely used ultraviolet (UV) filter in personal care products, through in silico analyses.Material and Method: The toxicological properties of BP-3 were predicted using the ProTox and deTOX platforms. Developmental toxicity was modeled specifically for the first, second, and third trimesters of pregnancy. Additionally, the potential interactions of BP-3 with human cytochrome P450 enzymes (CYP1A2, CYP2C19, CYP2D6) were investigated via molecular docking analyses.Result and Discussion: Literature data confirm that BP-3 can cross the placental barrier and enter systemic circulation. In silico predictions indicated a high developmental toxicity potential, particularly during the second trimester (probability: 0.952). Molecular docking analyses revealed strong binding affinity of BP-3 to CYP1A2, as well as notable interactions with CYP2C19 and CYP2D6, suggesting possible interference with drug metabolism and endocrine regulation. The findings were further supported by structural toxicity contribution mapping, highlighting specific molecular regions responsible for toxicity. ## In Vitro and In Silico Assessment of the Anti-Cancer Potential of Pinus sylvestris and Picea abies Needle Essential Oils - Source: Applied Sciences (journals) - Date: 2026-09-05T00:00:00Z - Categories: Docking & Screening - Journal: Applied Sciences - DOI: 10.3390/app16178839 - External ID: 82d119a20030f38d4e182c1a271a6cd16dd0accb - Keywords: molecular docking, binding affinity - Source URL: - Dashboard article: Abstract: Cancer remains a leading cause of global mortality, entailing the exploration of novel therapeutic agents from natural sources. This study evaluated the in vitro anti-cancer potential of essential oils (EOs) extracted from the needles of Pinus sylvestris L. (Scots pine) and Picea abies (L.) H. Karst. (Norway spruce). GC–MS analysis revealed that both EOs are rich in monoterpenes; the primary constituents were α-pinene (35.27%) for Pinus sylvestris sylvestris EO (Pinus EO) and β-pinene (31.40%) for Picea abies EO (Picea EO). Cell viability assay (Alamar Blue) demonstrated that both EOs exert a dose-dependent cytotoxic effect across all tested malignant lines: melanoma (A375), colorectal adenocarcinoma (HT-29), and pancreatic adenocarcinoma (PANC-1). Notably, Pinus EO consistently displayed higher potency toward cancer cells when compared with normal keratinocytes (HaCaT), with the highest selectivity noticed in HT-29 cells. Immunofluorescence analysis revealed morphological changes consistent with apoptosis, such as nuclear condensation, DNA fragmentation, and the disruption of the β-actin cytoskeleton. High-resolution respirometry indicated that the EOs increase LEAK respiration, while also significantly reducing the oxidative phosphorylation (OXPHOS) efficiency. Furthermore, network pharmacology and molecular docking predicted EP300 as a common hub target for both EOs, with β-caryophyllene showing the most favourable predicted binding affinity among the investigated constituents. Overall, these preliminary in vitro and in silico findings support the further investigation of Pinus EO and Picea EO as potential candidates for the development of future adjuvant strategies in cancer therapy. ## Metabolic engineering and deep learning-driven protein engineering for N-Acetylneuraminic acid biosynthesis in Escherichia coli - Source: Nature Communications (journals) - Date: 2026-09-05T00:00:00+00:00 - Authors: Nan-Kai Wang, Song Yue, Jin-Ping Chen, Chang Su, Zhen-Ming Lu, Jin-Song Gong, Wei E. Huang, Zheng-Hong Xu, Jin-Song Shi - Journal: Nature Communications - DOI: 10.1038/s41467-026-77406-2 - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Prediction of Ocular Toxicity of Prostaglandin F2α Analogs Based on Local Computational Models: Superiority over Global Models and Experimental Validation - Source: Molecules (journals) - Date: 2026-09-05T00:00:00Z - Categories: Cheminformatics, Docking & Screening - Journal: Molecules - DOI: 10.3390/molecules31173116 - External ID: 4f7bc051d755656a915c25f8e4a2a97a266c00fb - Keywords: molecular fingerprints, molecular docking, IC50, molecular representations - Source URL: - Dashboard article: Abstract: Drug-induced ocular toxicity is difficult to predict and evaluate, particularly for prostaglandin F2α (PGF2α) analogs used in the management of glaucoma. Traditional global predictive models, which are built with large and various datasets (n = 6187), provide systematic false-negative results for groups of structurally uniform compounds. To address this limitation, we developed special local classification models for PGF2α analogs. A structurally consistent local training dataset (n = 350) was assembled using Murcko scaffold filtering and Tanimoto similarity selection (≥0.5). Binary classification models were built using three different molecular fingerprints in combination with machine learning and two deep learning methods. Although two-dimensional molecular representations do not encode stereochemistry, computational predictions still apply to the shared two-dimensional scaffold of these compounds. These improved models were used to predict the ocular toxicity of latanoprost and its related impurities. The local models (n = 350) provided accurate toxicity predictions for latanoprost, latanoprost acid, 15(S)-latanoprost, and the trans-5,6-latanoprost isomer, whereas all 16 global computational models provided false-negative results. The experimental assessment performed with primary rabbit corneal epithelial cells (pRCECs) and a human corneal epithelial cell line (HCE-T) confirmed the cytotoxic effects, which were in agreement with the predictions made by the models. Among the tested compounds, 15(S)-latanoprost exhibited the highest cytotoxicity (IC50 = 86.22 μM, 95% CI: 82.90–89.56 μM), followed by trans-5,6-latanoprost (IC50 = 106.70 μM, 95% CI: 103.4–110.0 μM) and latanoprost (IC50 = 112.60 μM, 95% CI: 106.7–118.6 μM). At the standard therapeutic dosage (0.005%), no significant toxic response was observed. Virtual molecular docking was employed to explore the mechanism, and all analogs docked favorably into the quinone-binding channel of mitochondrial complex I and the catalytic cleft of SIRT3. These results show how a localized modeling approach is better able to capture structure–toxicity correlations among chemically similar compounds and highlight the critical need for rigorous impurity management in latanoprost products, particularly regarding 15(S)-latanoprost. ## Quercetin attenuates emamectin benzoate-induced myocardial PANoptosis in chickens: Critical role of the MAPK/NF-κB signaling pathway and mitochondrial dynamics. - Source: International immunopharmacology (journals) - Date: 2026-09-05T00:00:00Z - Categories: Docking & Screening - Journal: International immunopharmacology - DOI: 10.1016/j.intimp.2026.117349 - External ID: f58d1a38ecf2d77c2a4069fee11e76bd234cb135 - Keywords: Molecular docking - Source URL: - Dashboard article: Abstract: Emamectin benzoate (EMB) is a highly effective and widely detected environmental insecticide whose potential to induce severe cardiotoxicity and localized myocardial inflammation has raised substantial biomedical concerns. Quercetin (Que), a ubiquitous plant-derived natural antioxidant, has shown promise in mitigating exogenous chemical-induced toxicities, yet its precise immunopharmacological mechanism against EMB-induced cardiotoxicity remains to be fully elucidated. This study investigated the protective effects and mechanisms of Que. against EMB-induced cardiotoxicity by integrating network toxicology, transcriptomics, and in vivo/in vitro experimental validation. Multi-omics analysis identified MAPK8, MAPK3, and CASP3 as key responsive targets conserved across species. Functional enrichment revealed that these targets are predominantly involved in the MAPK/NF-κB signaling pathway and mitochondrial organization. Molecular docking and targeted intervention experiments confirmed that Que. potentially targets and binds to JNK (encoded by MAPK8), thereby blocking the EMB-activated signaling axis. This molecular interaction effectively restored mitochondrial dynamics and prevented mitochondrial DNA leakage into the cytoplasm, ultimately alleviating myocardial PANoptosis. Collectively, our findings demonstrate that Que. serves as a natural antagonist against chemical-induced cardiotoxicity by modulating the MAPK/NF-κB axis, offering a promising therapeutic strategy for mitigating pesticide-related immunotoxicological risks. ## The Potential of Bioactive Compounds from Abelmoschus manihot (L.) Medik Leaf Againts Acne Vulgaris through Network Pharmacology and Molecular Docking - Source: Bulletin of Pharmaceutical Sciences Assiut University (journals) - Date: 2026-09-05T00:00:00Z - Categories: Docking & Screening - Journal: Bulletin of Pharmaceutical Sciences Assiut University - DOI: 10.21608/bfsa.2026.517358.3247 - External ID: 22125c6b19e2f61a8e3425530e71ff19ff695e9b - Keywords: Molecular Docking - Source URL: - Dashboard article: - Abstract: not stored for this record. ## Thermal conductivity predictions with foundation atomistic models - Source: Nature Communications (journals) - Date: 2026-09-05T00:00:00+00:00 - Authors: Balázs Póta, Paramvir Ahlawat, Gábor Csányi, Michele Simoncelli - Journal: Nature Communications - DOI: 10.1038/s41467-026-76391-w - Source URL: - Dashboard article: Abstract: Advances in machine learning have led to the development of foundation models for atomistic materials chemistry, enabling quantum-accurate descriptions of interatomic forces across chemically diverse compounds at reduced computational cost. Hitherto, the accuracy and utility of these models have been assessed relying on descriptors based on formation energies or idealized harmonic atomic vibrations. Yet, the rigorous and physically interpretable quantification of their capability to describe both realistic anharmonic atomic dynamics and technologically relevant observables remains a pressing problem. Here, we address this problem, leveraging the Wigner formulation of heat transport and the Grüneisen approach to thermal expansion to connect the atomic-physics awareness of foundation models to their utility in predicting experimentally observable thermomechanical properties, presenting standards and fine-tuning protocols needed to achieve first-principles accuracy. We apply our framework to a database of 103 solids with diverse compositions and structures, demonstrating that it overcomes the major bottlenecks of current methods for designing heat-management materials — high cost, limited transferability, or lack of physics awareness — and has the potential to discover materials for technologies ranging from thermal insulation to neuromorphic computing. ## Thermodynamics and Unbinding Kinetics of A22 at Multiple Actin-Binding Sites Revealed by Enhanced Sampling Simulations - Source: Journal of Chemical Information and Modeling (journals) - Date: 2026-09-05T00:00:00+00:00 - Categories: Docking & Screening, Free Energy & MD - Authors: Anuj Kumar, Debabrata Pramanik - Journal: Journal of Chemical Information and Modeling - DOI: 10.1021/acs.jcim.6c01530 - Keywords: molecular docking, binding affinity, molecular dynamics, MD simulations - Source URL: - Dashboard article: Abstract: The cytoskeletal protein plays a major role in various cellular processes. Understanding the interactions of small molecules with cytoskeletal protein therefore might help in the development of therapeutics. Employing combined molecular docking, all-atom molecular dynamics (MD), and enhanced sampling technique, we studied bacterial inhibitor A22 and actin protein interaction. Five probable A22 binding sites (S1–S5) were observed in actin and unbiased MD simulations of 100 ns to 1000 ns showed structural stability at these sites. Interaction analyses showed A22 to form transient interactions except at sites S1, S3, and S5 where long-lived interactions were present. Enhanced sampling simulations quantitatively estimated ligand dissociation free energy (ΔGb) ∼ −1.3 ± 0.6 kcal/mol to −4.9 ± 1.5 kcal/mol for sites S1–S4 with residence times ∼μs to ms. For site S5, we observed the highest binding affinity with dissociation free energy (ΔGb) ∼ −8.8 ± 3.8 kcal/mol with residence time ∼sec. Analyses of the dissociation trajectories predict multiple dissociation pathways for A22 and found key gatekeeper residues at specific-sites facilitating ligand dissociation. Comparison of A22 with other known inhibitors suggests that A22 binds actin with relatively lower affinity, however, exhibits site-specific unbinding kinetics. Thus, our study provides a detailed mechanistic overview of actin + A22 interaction, its various binding modes, binding affinities, and unbinding kinetics. It also shows the importance and usage of metadynamics and its variants in exploring rare events like ligand–protein interaction. Deeper insights gained from this study expands our understanding of cytoskeletal ligand dynamics. These knowledges will be paramount in designing drug targeting cytoskeletal protein actin. ## Worenine attenuates pulmonary fibrosis potentially through modulation of PBK/SRC-associated TGF-β1/Smad and NF-κB signaling pathways. - Source: Biomedicine & pharmacotherapy = Biomedecine & pharmacotherapie (journals) - Date: 2026-09-05T00:00:00Z - Categories: Docking & Screening - Journal: Biomedicine & pharmacotherapy = Biomedecine & pharmacotherapie - DOI: 10.1016/j.biopha.2026.119868 - External ID: 2a5b0ca254a1c2ca0dae27eb608a38f33074bf7c - Keywords: molecular docking - Source URL: - Dashboard article: Abstract: OBJECTIVE To elucidate the pharmacological mechanisms of worenine in pulmonary fibrosis (PF) through an integrative strategy combining network pharmacology, molecular docking, and experimental validation. METHODS Potential targets of worenine and PF-related genes were obtained from public databases and integrated to construct interaction networks. Protein-protein interaction (PPI), GO, and KEGG enrichment analyses were performed to identify key targets and pathways. Molecular docking evaluated binding affinities between worenine and hub targets. The anti-fibrotic effects and mechanisms of worenine were validated in bleomycin-induced PF mice and in TGF-β1-stimulated A549 and MRC-5 cells. RESULTS A total of 116 overlapping targets were identified. Key targets included SRC, IL6, TNF, NFKB1, and PIK3CA. Enrichment analyses indicated that worenine regulates pathways related to inflammation and cellular stress, including TNF, IL-17, and HIF-1 signaling. Molecular docking showed strong binding affinities between worenine and core targets such as EZH2, PTGS2, PBK, SRC, NFKB1, and IL6. In vivo, worenine alleviated PF, reducing collagen deposition and improving histopathology. In vitro, worenine inhibited EMT and FMT, accompanied by suppression of PBK/SRC-associated TGF-β1/Smad, ERK, and NF-κB signaling, as well as decreased secretion of IL-6, IL-8, and VEGFA. CONCLUSION Worenine attenuates experimental pulmonary fibrosis, potentially through modulation of PBK/SRC-associated signaling and the TGF-β1/Smad, ERK, and NF-κB pathways. These findings provide a rationale for further investigation of Worenine as a potential therapeutic candidate for PF.