Journal of Chemical Information and Modeling 最新文献

{"title":"Integrating Chemo- and Bioinformatics with In Vitro Biological Assays to Discover Potential ACE2 and Mpro Inhibitors against SARS-CoV-2.","authors":"Ryan S Ramos,João S N de Souza,Mariana H Chaves,Joaquín M Campos,Willyenne M Dantas,Lindomar J Pena,Maracy L D S Andrade,Cleydson B R Santos","doi":"10.1021/acs.jcim.5c01056","DOIUrl":"https://doi.org/10.1021/acs.jcim.5c01056","url":null,"abstract":"The study aims to identify potential SARS-CoV-2 inhibitors and investigate the mechanism of action on the viral ACE2 receptor and main protease (Mpro), using chemo- and bioinformatics approaches. Ligand-based virtual screening was performed in the Molport database (∼4.79 million compounds), and after applying physicochemical filters, 313 molecules with characteristics such as hydroxychloroquine were obtained. After obtaining bioactive conformations, the molecular structures were subjected to the study of pharmacokinetic predictions, in which 106 molecules presented properties for oral bioavailability, penetration of the BBB, PPB, and solubility (average). The toxicological property predictions proved plausible for the molecules, as they did not present warnings of hepatotoxicity, mutagenicity, potential risk of carcinogenicity, and LC50 and LD50 values higher than the controls. Subsequently, 81 structures were subjected to a molecular docking study of ACE2 receptor/Spike and Mpro. In the ACE2 receptor, four (4) ligands showed high binding affinity value, in which the molecule MolPort-010-778-422 had the best ΔG value of -9.414 kcal/mol, followed by MolPort-009-093-282 with ΔG = -8.978 kcal/mol. In the Mpro receptor, four (4) ligands showed high binding affinity values compared to control 11b, with emphasis on molecule MolPort-005-766-143 with ΔG = -8.829 kcal/mol, followed by molecule MolPort-046-186-743. To study the antiviral effects of the molecules in vitro, TopHits8 molecules were tested against the SARS-CoV-2 virus. MolPort-010-778-422 had the best result on the screening and presented an IC50 of 8.9 nM.","PeriodicalId":44,"journal":{"name":"Journal of Chemical Information and Modeling ","volume":"12 1","pages":""},"PeriodicalIF":5.6,"publicationDate":"2025-07-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144701378","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Machine Learning-Driven Prediction of Electrochemical Promotion in the Reverse Water Gas Shift Reaction.","authors":"Ju Wang,Hongying Zhou,Mustapha Ezzeddine,Karim Harb,Sayed Ahmed Ebrahim,Elena A Baranova","doi":"10.1021/acs.jcim.5c00927","DOIUrl":"https://doi.org/10.1021/acs.jcim.5c00927","url":null,"abstract":"Electrochemical promotion of catalysis (EPOC) provides an effective and versatile strategy to enhance catalytic activity, selectivity, and stability in the reverse water-gas shift (RWGS) reaction, facilitating efficient CO2 hydrogenation to syngas under milder conditions. However, predicting EPOC results using novel catalytic materials under diverse conditions remains challenging. This study introduces a machine learning framework to predict electrochemical promotion behavior and the rate enhancement ratios (ρ), i.e., ratio between promoted and unpromoted reaction rate, based on the selected catalyst, reaction, and electrochemical condition descriptors. Several classification and regression models were trained and tested using a data set compiled from previous studies. The best-performing random forest (RF) and extreme gradient boosting (XGB) models were validated with new experimental data collected from systems employing lithium lanthanum titanate (LLTO) solid electrolyte and Pt-ZnO catalysts, achieving an R2 of 0.97 and a mean squared error (MSE) of 0.01. This data-driven approach is interpretable, generalizable to other catalytic systems, and provides a powerful tool for advancing the development of catalytic materials for EPOC in RWGS reactions.","PeriodicalId":44,"journal":{"name":"Journal of Chemical Information and Modeling ","volume":"25 1","pages":""},"PeriodicalIF":5.6,"publicationDate":"2025-07-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144701376","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"MMF-MCP: A Deep Transfer Learning Model Based on Multimodal Information Fusion for Molecular Feature Extraction and Carcinogenicity Prediction.","authors":"Liwei Liu,Qi Zhang,Yuxiao Wei","doi":"10.1021/acs.jcim.5c01362","DOIUrl":"https://doi.org/10.1021/acs.jcim.5c01362","url":null,"abstract":"Molecular carcinogenicity is a crucial factor in the development of cancer, and accurate prediction of it is vital for cancer prevention, treatment, and drug development. In recent years, deep learning has been applied to predict molecular carcinogenicity, but due to limitations in data quality and feature richness, these methods still need improvement in terms of accuracy, robustness, and interpretability. In this article, we propose a deep transfer learning model based on multimodal information fusion, called MMF-MCP, for molecular feature extraction and carcinogenicity prediction. We extract molecular graph features and fingerprint features using graph attention networks and convolutional neural networks, respectively, and process molecular images through a deep residual network, SE-ResNet18, equipped with a squeeze-and-excitation module. To more effectively utilize limited carcinogenicity data and enhance the model's predictive performance and generalization ability, we further apply a transfer learning strategy by pretraining the model on a molecular mutagenicity data set and then fine-tuning it on the carcinogenicity data set, enabling knowledge transfer and significant improvement in model performance. MMF-MCP achieves average ACC, AUC, SE, and SP scores of 0.8452, 0.8513, 0.8571, and 0.8333 on benchmark data sets for molecular carcinogenicity, significantly outperforming state-of-the-art molecular carcinogenicity prediction methods. Additionally, the visualization results of MMF-MCP on molecular images demonstrate its strong interpretability, providing significant assistance in visually observing and understanding the critical structures and features of molecular carcinogenicity. The source code for MMF-MCP is available at https://github.com/liuliwei1980/MCP.","PeriodicalId":44,"journal":{"name":"Journal of Chemical Information and Modeling ","volume":"20 1","pages":""},"PeriodicalIF":5.6,"publicationDate":"2025-07-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144701381","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Benchmarking 3D Structure-Based Molecule Generators.","authors":"Natasha Sanjrani, Damien E Coupry, Peter Pogány, David S Palmer, Stephen D Pickett","doi":"10.1021/acs.jcim.5c01020","DOIUrl":"https://doi.org/10.1021/acs.jcim.5c01020","url":null,"abstract":"<p><p>To understand the benefits and drawbacks of 3D combinatorial and deep learning generators, a novel benchmark was created focusing on the recreation of important protein-ligand interactions and 3D ligand conformations. Using the BindingMOAD data set with a hold-out blind set, the sequential graph neural network generators, Pocket2Mol and PocketFlow, diffusion models, DiffSBDD and MolSnapper, and combinatorial genetic algorithms, AutoGrow4 and LigBuilderV3, were evaluated. It was discovered that deep learning methods fail to generate structurally valid molecules and 3D conformations, whereas combinatorial methods are slow and generate molecules that are prone to failing 2D MOSES filters. The results from this evaluation guide us toward improving deep learning structure-based generators by placing higher importance on structural validity, 3D ligand conformations, and recreation of important known active site interactions. This benchmark should be used to understand the limitations of future combinatorial and deep learning generators. The package is freely available under an Apache 2.0 license at github.com/gskcheminformatics/SBDD-benchmarking.</p>","PeriodicalId":44,"journal":{"name":"Journal of Chemical Information and Modeling ","volume":" ","pages":""},"PeriodicalIF":5.6,"publicationDate":"2025-07-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144715006","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Augmenting MACCS Keys with Persistent Homology Fingerprints for Protein-Ligand Binding Classification.","authors":"Johnathan W Campbell, Konstantinos D Vogiatzis","doi":"10.1021/acs.jcim.5c00934","DOIUrl":"https://doi.org/10.1021/acs.jcim.5c00934","url":null,"abstract":"<p><p>Machine learning has become an essential tool in computational drug design, enabling models to uncover patterns in molecular data and predict protein-ligand interactions. This study introduces a novel approach by integrating persistence images with MACCS Keys to construct a more robust and enriched molecular representation. By incorporating topological descriptors that capture the intrinsic geometry and connectivity of the molecular structure, we aim to enhance classification performance by providing complementary information to common cheminformatic fingerprints. Using a consistent artificial neural network architecture and training setup, we evaluate this approach across 19 protein-ligand bioactivity datasets available from ChEMBL. We generate persistence images using topological data analysis and concatenate them with MACCS Keys. Our results demonstrate that this augmented representation consistently outperforms its components, yielding a higher average validation Matthews correlation coefficient across all but one dataset. These findings highlight the potential of integrating molecular shape-based features with traditional descriptors to enhance predictive performance for computer-aided drug design workflows.</p>","PeriodicalId":44,"journal":{"name":"Journal of Chemical Information and Modeling ","volume":" ","pages":""},"PeriodicalIF":5.6,"publicationDate":"2025-07-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144717058","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Comparative Analysis of Polarizable and Nonpolarizable CHARMM Family Force Fields for Proteins with Flexible Loops and High Charge Density.","authors":"Sangram Prusty,Rafael Brüschweiler,Qiang Cui","doi":"10.1021/acs.jcim.5c01328","DOIUrl":"https://doi.org/10.1021/acs.jcim.5c01328","url":null,"abstract":"Electrostatic interactions are fundamental to biomolecular structure, stability, and function. While these interactions are traditionally modeled using fixed-charge force fields, such approaches are not transferable among different molecular environments. Polarizable force fields, such as DRUDE, address this limitation by explicitly incorporating the polarization effect. However, their performance does not uniformly surpass that of nonpolarizable force fields since multiple factors such as bonded terms, dihedral correction maps, and solvent screening also modulate biomolecular dynamics. In this work, we study the Im7 protein to evaluate the structural and dynamic behaviors of nonpolarizable (CHARMM36m) and polarizable (DRUDE2019) force fields relative to NMR experiments. Our simulations show that DRUDE2019 better stabilizes α-helices than CHARMM36m, including shorter ones that contain helix-breaking residues. However, both force fields underestimate loop dynamics, particularly in the loop I region, mainly due to restricted dihedral angle sampling. Moreover, salt bridge analysis reveals that DRUDE2019 and CHARMM36m preferentially stabilize different salt bridges, driven by ionic interactions, charge screening by the environment, and neighboring residue flexibility Additionally, the latest DRUDE2019 variant, featuring updated NBFIX and NBTHOLE parameters for ion-protein interactions, demonstrated improved accuracy in modeling Na+-protein interactions. These findings are further supported by simulations of CBD1, a protein with a β-sheet and flexible loops, which exhibited similar trends of stable structured regions and restricted loop dynamics across both force fields. These findings highlight the need to balance bonded and nonbonded interactions along with dihedral correction maps while incorporating polarization effects to improve the accuracy of force fields to model protein structure and dynamics.","PeriodicalId":44,"journal":{"name":"Journal of Chemical Information and Modeling ","volume":"53 1","pages":""},"PeriodicalIF":5.6,"publicationDate":"2025-07-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144693192","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Expert-Guided Substructure Information Bottleneck for Molecular Property Prediction.","authors":"Tianyi Jiang,Qiang Yao,Zeyu Wang,Xiaoze Bao,Shanqing Yu,Qi Xuan","doi":"10.1021/acs.jcim.5c00456","DOIUrl":"https://doi.org/10.1021/acs.jcim.5c00456","url":null,"abstract":"Molecular property prediction plays a crucial role in cheminformatics, yet existing methods are constrained by data scarcity and molecular structural heterogeneity. The Mixture of Experts (MoE) framework adopts a divide-and-conquer approach by partitioning the input space and employing expert models. However, current methods primarily rely on scaffold or atomic-level information, often neglecting fine-grained features such as functional groups. Moreover, existing MoE models lack effective mechanisms to filter redundant and noisy information, limiting prediction accuracy and generalization. To address these challenges, we propose a novel Expert-Guided Substructure Information Bottleneck (ESIB-Mol) framework that integrates MoE learning with the Information Bottleneck (IB) principle to optimize molecular representation learning. ESIB-Mol employs substructure-specific experts to focus on key molecular scaffolds and functional groups, which play a crucial role in determining molecular properties such as bioactivity and pharmacokinetics. Meanwhile, the IB principle is leveraged to filter out redundant and irrelevant information, thereby enhancing prediction accuracy and interpretability. Additionally, a dynamic gating mechanism adaptively assigns molecules to the most relevant expert, optimizing computational efficiency. Extensive experiments on benchmark data sets demonstrate the effectiveness of ESIB-Mol, highlighting its superior performance in molecular property prediction.","PeriodicalId":44,"journal":{"name":"Journal of Chemical Information and Modeling ","volume":"71 1","pages":""},"PeriodicalIF":5.6,"publicationDate":"2025-07-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144693193","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A Computationally Efficient Method to Generate Plausible Conformers for Ensemble Docking and Binding Free Energy Calculations.","authors":"Ö Zeynep Güner Yılmaz,Pemra Doruker,Ozge Kurkcuoglu","doi":"10.1021/acs.jcim.5c00431","DOIUrl":"https://doi.org/10.1021/acs.jcim.5c00431","url":null,"abstract":"This study presents a computationally efficient approach to generate plausible protein conformers for ensemble docking to enable evaluations of interactions between ligand and protein for ranking the docked ligands according to their binding affinities. Two binding regions of triose phosphate isomerase (TIM), its catalytic site with DHAP (G. gallus TIM), and its dimer interface with 3PG (P. falciparumTIM) involving flexible loops were investigated as case studies. The binding sites of the apo and holo forms were modeled at the atomistic scale (high resolution) while the remaining structure was coarse-grained (low resolution) leading to a mixed-resolution description of the protein. The slowest three normal modes related to the functional dynamics of TIM were obtained using the Anisotropic Network Model and employed to derive 36 conformers of the truncated high-resolution regions by assessing six deformation parameters in both directions of the harmonic motions. Through energy minimization and docking calculations in Glide, optimal extents of deformation were identified. The docked truncated structures were then subjected to independent molecular dynamics (MD) simulations to confirm the interactions of the ligands in the binding sites. To prevent the disintegration of the truncated structure, different buffer zones and harmonic restraints were assessed to finally decide on four distinct zones with restraints of 0, 25, 35, and 50 kcal/mol·Å2. Each conformer underwent 900 ns-long simulations across three replicates reaching a total simulation time of 15.2 μs. Binding free energy calculations were conducted using the MM-GBSA approach using the first 50, first 100, first 200, and 300 ns intervals, which pointed out that 100 ns-long simulations were sufficient to estimate the binding affinities for TIM. Results consistently indicated comparable binding energies between the intact and truncated TIM structures underscoring the approach's reliability, where the truncated conformers also offered varying binding site geometries yielding favorable interactions. Comparative docking at the dimer interface of G. gallus and P. falciparum TIM further highlighted species-specific binding dynamics, affirming the methodology's applicability for diverse biological questions and establishing a computationally efficient approach to estimate binding free energy values even for supramolecular assemblages.","PeriodicalId":44,"journal":{"name":"Journal of Chemical Information and Modeling ","volume":"678 1","pages":""},"PeriodicalIF":5.6,"publicationDate":"2025-07-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144684325","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Mechanistic Insights into the Binding of Different Antagonists to 5-HT1AR: A Molecular Docking and Molecular Dynamics Simulation Study.","authors":"Lulu Guan,Dushuo Feng,Jingxuan Ge,Bote Qi,Yunxiang Sun,Yu Zou","doi":"10.1021/acs.jcim.5c01240","DOIUrl":"https://doi.org/10.1021/acs.jcim.5c01240","url":null,"abstract":"The serotonin 1A receptor (5-HT1AR) is involved in a wide range of physiological processes, and it has attracted considerable attention as an important target for developing new medicines. The antagonists of 5-HT1AR display therapeutic potential in depression and cognitive dysfunction, and the elucidation of their interaction with the receptor is crucial to understand pharmacological actions and develop novel therapeutic agents. Herein, we performed conventional molecular docking and molecular dynamics (MD) simulations to address the docking pose and binding mechanisms of six representative antagonists (way100635, way101405, lecozotan, nan190, sdz216-525, and nad299) to 5-HT1AR. We found that among the six antagonists, sdz216-525 exhibited the most negative docking score at -9.5 kcal/mol, while nad299 displayed the least negative score. The common pharmacophore aromatic group appeared in all six antagonists, and piperazine existed in the five antagonists except for nad299. MD simulation results showed that upon the addition of antagonists, the conformation of 5-HT1AR was changed to various extents, and the relative positions of transmembrane 3 (TM3), TM5, and TM6 underwent rearrangement. Among these antagonists, lecozotan exhibited the highest binding affinity to 5-HT1AR, whereas nad299 showed the weakest interaction. The molecular recognition of six antagonists by 5-HT1AR involved different binding patterns, with variable contributions from hydrophobic, H-bonding, cation-π/anion-π, and aromatic stacking interactions. Taken together, our computational study contributes to the understanding of the binding mechanism of antagonists to 5-HT1AR, which may facilitate the design of new antagonists targeting 5-HT1AR.","PeriodicalId":44,"journal":{"name":"Journal of Chemical Information and Modeling ","volume":"115 1","pages":""},"PeriodicalIF":5.6,"publicationDate":"2025-07-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144684321","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"m5U-HybridNet: Integrating an RNA Foundation Model with CNN Features for Accurate Prediction of 5-Methyluridine Modification Sites.","authors":"Xinyu Li,Zhenjie Luo,Jingwei Lv,Chao Yang,Shankai Yan,Junlin Xu,Yajie Meng,Leyi Wei,Zilong Zhang,Quan Zou,Feifei Cui","doi":"10.1021/acs.jcim.5c01237","DOIUrl":"https://doi.org/10.1021/acs.jcim.5c01237","url":null,"abstract":"The 5-methyluridine (m5U) modification in RNA is vital for numerous biological processes, making its precise identification a key focus in computational biology. However, traditional wet-lab detection methods are cumbersome and time-consuming, whereas existing machine learning and deep learning computational prediction models still have room for improvement. Consequently, this study introduces m5U-HybridNet, an innovative framework that strategically integrates an RNA foundation model (RNA-FM) for deep semantic feature extraction with convolutional neural network-derived characteristics, attaining unparalleled success in identifying RNA m5U modification sites. Simultaneously, when compared with other existing models across different cell types and experimental techniques, it exhibits outstanding generalization capabilities. The m5U-HybridNet web server, accessible at http://www.bioai-lab.com/m5U, offers an effective and reliable platform for predicting RNA modification sites. It not only implies the diverse potential applications of pretrained models in the analysis of biological sequences but also enhances the application of data-driven machine intelligence in the realm of molecular biophysics principles.","PeriodicalId":44,"journal":{"name":"Journal of Chemical Information and Modeling ","volume":"18 1","pages":""},"PeriodicalIF":5.6,"publicationDate":"2025-07-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144684324","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}