Journal of Chemical Information and Modeling 最新文献

{"title":"ML-RKIM: Physics-Informed Machine Learning for Reaction Kinetics Identification and Modeling","authors":"Paulina Portales Picazo,&nbsp;Navid Zobeiry","doi":"10.1021/acs.jcim.6c01936","DOIUrl":"10.1021/acs.jcim.6c01936","url":null,"abstract":"<p>Kinetic processesplay a central role in structural transformationsacross natural, engineered, and biological systems, ultimately determiningtheir final properties. Traditional kinetic characterization methodsare often time-consuming and resource-intensive, and require priorassumptions about the reaction mechanism. In this work, we presenta physics-informed machine learning framework to extract closed-formkinetic representations from time-series data without preselectinga single reaction mechanism. The framework combines neural networks,physics-informed features, and sparse regularization. The frameworkwas applied to a range of processes, including thermally driven reactionssuch as polymer curing and crystallization, and nonthermal processessuch as enzymatic browning. The method identified sparse, closed-formkinetic representations while reducing reliance on a single predefinedkinetic model. R² values exceeded 98% for the syntheticcases and ranged from 76% to 92% for the experimental cases. In addition,a systematic ablation study evaluated robustness to noise, data availability,sampling density, incomplete measurements, and candidate functionlibrary variation, and an experimental case study was included toassess the performance under realistic conditions. These results demonstratethe potential of this approach to accelerate kinetic characterizationacross polymer, food, and biomolecular processing applications, whereaccurate kinetic models are essential.</p>","PeriodicalId":44,"journal":{"name":"Journal of Chemical Information and Modeling ","volume":"66 19","pages":"12448–12462"},"PeriodicalIF":6.4,"publicationDate":"2026-10-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149021790","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 Experiments with External Data, Domain-Specific Features, and Learning Strategies for Modeling ADME and Potency of Coronavirus Main Protease Inhibitors","authors":"Ida Titus,&nbsp;Ashok Palaniappan","doi":"10.1021/acs.jcim.6c00304","DOIUrl":"10.1021/acs.jcim.6c00304","url":null,"abstract":"<p>Predictive computational models of potency and pharmacokineticproperties of antiviral drug candidates can dramatically acceleratethe path from molecular design to clinical evaluation. The ASAP-Polaris-OpenADMETconsortium hosted the open-science Antiviral Drug Discovery Challenge2025, providing a rigorous setting for this problem and requiringprediction of potency against the SARS-CoV-2 and MERS-CoV <em>M</em>^<em>pro</em> alongside five ADME endpoints (LogD, KSOL, HLM, MLM, and MDR1-MDCKII). In response, we pursueda domain-aware modeling strategy, reasoning that the distinctive chemicalproperties of antiviral compounds could be explicitly engineered inthe feature space. We tested this premise by isolating the key moleculardescriptors capable of identifying antivirals, augmenting challengedata sets with curated public data, and building both classical machinelearning and graph neural network (GNN) variants. For ADME modeling,ensemble gradient boosting models trained on our bespoke feature spaceconsistently outperformed both broader tree-based approaches and GNNvariants across the five ADME end points in blind evaluation. By afactorial design of experiments and statistical benchmarking, we evaluatedthe performance of the optimal XGBoost model for each end point. Formodeling potency, we explored varied pretraining paradigms and ensemblefusion strategies, and found that a stacking ensemble combining docking-enhancedXGBoost with a stereochemistry-aware AttentiveFP Graph-Transformermodel yielded the strongest significant generalization on the Challengetest set, suggesting that orthogonal model families reduced epistemicuncertainty attributable to model selection in data-limited, multitargetsettings. Beyond benchmarking, our experiments provided a testbedfor evaluating critical assumptions in the field, encompassing dataaugmentation strategies, splitting protocols, and ensemble integrationapproaches. Our analysis revealed that a consistently optimal modelis elusive, and that several standard industry practices fail to withstandclose inspection. Overall, our work provides both practical benchmarkedmodeling strategies for ADME and potency prediction and a reproduciblecomputational foundation for AI-driven antiviral drug discovery efforts.</p>","PeriodicalId":44,"journal":{"name":"Journal of Chemical Information and Modeling ","volume":"66 19","pages":"12549–12567"},"PeriodicalIF":6.4,"publicationDate":"2026-10-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149090891","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":"T-ANALYZE as a Workflow for Matched Molecule Pair Analysis on Peptides","authors":"Robert P. Sheridan","doi":"10.1021/acs.jcim.6c01414","DOIUrl":"10.1021/acs.jcim.6c01414","url":null,"abstract":"<p>Matchedmolecular pairs (MMP) is an approach to finding local changesin structure that give rise to consistent changes in activity. Assumingmolecules A and B are in a matched pair, one can imagine a “transformation”,i.e., adding, removing, or exchanging a small set of atoms that changesA to B. Typical machine-learning algorithms are meant to handle individualmolecules A, and model the activity difference between A and the averageactivity, and are agnostic to the presence of similar compounds inthe training set. In contrast, MMP looks for differences between similarcompounds A and B and is agnostic to the absolute value of the activitiesof A and B. Machine learning models tend to be poor at predictingactivity cliffs, while MMP is designed to identify them. Previously,we (Sheridan et al., <em>J. Chem. Inf. Model.</em> <strong>2006</strong>, 46, 180–192) suggested a workflow T-ANALYZE that organizesand displays sets of closely related drug-sized compounds such thatany consistent MMP-based structure-activity in a data set can be easilyseen. Nowadays, peptides are under consideration as candidate drugsand the MMP approach should be extended to them. Peptides are smallenough that they can be treated as connection tables, i.e., atomsconnected with bonds. Alternatively, they can be written as sequences,i.e., a one-dimensional list of monomer names. In this paper, we adaptthe T-ANALYZE workflow to peptides represented both as connectiontables and as sequences. We demonstrate this on the PAMPA activityof the CycPeptMPDB dataset.</p>","PeriodicalId":44,"journal":{"name":"Journal of Chemical Information and Modeling ","volume":"66 19","pages":"12624–12636"},"PeriodicalIF":6.4,"publicationDate":"2026-10-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149090896","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":"AdventML: Advanced Enzyme Temperature Prediction with Transformer-Based Embeddings and Resampling Strategies","authors":"Jaldert François,Bart De Moor,Vera van Noort","doi":"10.1021/acs.jcim.6c02109","DOIUrl":"https://doi.org/10.1021/acs.jcim.6c02109","url":null,"abstract":"Accurate prediction of enzyme optimal catalytic temperature (Topt) is important for enzyme applications, particularly for identifying biocatalysts stable in extreme conditions, but experimental measurement remains costly and time-consuming. Here, we present AdventML, a machine learning framework for (Topt) prediction that expands the available models, evaluates protein representations and model architectures, and compares resampling strategies for improving performance at extremes. The base AdventML model achieved the best global held-out performance among the evaluated methods (R2 ≈ 0.64), MAE ≈7–8 °C, while tail-focused variants improved selected extreme-temperature regimes, reducing tail MAE by up to ∼1.8 °C. AdventML also showed that competitive (Topt) prediction is possible in the absence of organism growth temperature. Our Topt prediction models are made freely available as AdventML on GitHub.","PeriodicalId":44,"journal":{"name":"Journal of Chemical Information and Modeling ","volume":"37 1","pages":""},"PeriodicalIF":5.6,"publicationDate":"2026-10-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149152068","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":"MARS-RNA: An Automated Pipeline for Correlating RNA Molecular Dynamics Trajectories with Chemical Probing Data","authors":"Saeid Ekrami,&nbsp;Elisa Frezza","doi":"10.1021/acs.jcim.6c03140","DOIUrl":"10.1021/acs.jcim.6c03140","url":null,"abstract":"<p>Ribonucleic acid(RNA) molecules fold into complex, highly dynamicthree-dimensional conformational structures that govern their biologicalactivities. Chemical probing methods, such as SHAPE (selective 2′-hydroxylacylation analyzed by primer extension), provide structural insightsin the form of one-dimensional (1D) nucleotide reactivity profiles.However, mapping these one-dimensional profiles back to the underlyingdynamic structures remains a key challenge, as traditional analysesrely on a static view based on three-dimensional (3D) high-resolutionstructures from X-ray crystallography or nuclear magnetic resonance,rather than accounting for the dynamic nature of the system. In thisApplication Note, we introduce <span>MARS-RNA</span> (Moleculardynamics Analysis for Reactivity and Structure), an automated pipelinethat extracts and integrates multiscale structural and geometric featuresfrom RNA Molecular Dynamics (MD) trajectories, combining them withsecondary (2D) and 3D annotations from X3DNA-DSSR to identify featuresthat correlate with experimental reactivity. <span>MARS-RNA</span> processes raw GROMACS trajectories, performs frame-by-frame structuralannotation, calculates spatial centroids and physical distance fluctuations,computes Pearson/Spearman correlation coefficients against SHAPE data,and preprocesses results for machine learning pipelines. We show itsvalue on the RNA 3′-UTR (PDB ID: 1AUD, chain A), by highlighting several singlecorrelation with SHAPE reactivity.</p>","PeriodicalId":44,"journal":{"name":"Journal of Chemical Information and Modeling ","volume":"66 19","pages":"12224–12230"},"PeriodicalIF":6.4,"publicationDate":"2026-10-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149049608","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":"Correction to “PoseidonQ: A Free Machine Learning Platform for the Development, Analysis, and Validation of Efficient and Portable QSAR Models for Drug Discovery”","authors":"Muzammil Kabier,&nbsp;Nicola Gambacorta,&nbsp;Fulvio Ciriaco,&nbsp;Fabrizio Mastrolorito,&nbsp;Sunil Kumar,&nbsp;Bijo Mathew,&nbsp;Orazio Nicolotti","doi":"10.1021/acs.jcim.6c03051","DOIUrl":"10.1021/acs.jcim.6c03051","url":null,"abstract":"","PeriodicalId":44,"journal":{"name":"Journal of Chemical Information and Modeling ","volume":"66 19","pages":"13018–13018"},"PeriodicalIF":6.4,"publicationDate":"2026-10-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149049613","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":"Targeting the “Undruggable” STAT3 DBD: Free-Energy Perturbation-Guided Design of Covalent Micheliolide Derivatives Targeting Cysteine 367","authors":"Yangping Deng,&nbsp;Hengwei Bian,&nbsp;Fengyuan Zhang,&nbsp;Jianping Lin,&nbsp;Hongbo Li,&nbsp;Yue Chen,&nbsp;Jing Li,&nbsp;Jianshuang Guo,&nbsp;Haohao Fu","doi":"10.1021/acs.jcim.6c03067","DOIUrl":"10.1021/acs.jcim.6c03067","url":null,"abstract":"<p>Pancreatic cancer remains a highly lethal malignancy,largely dueto persistent STAT3 signaling that sustains cancer stem cells (CSCs)and drives therapeutic resistance. Conventional inhibitors targetingthe conserved SH2 domain of STAT3 have exhibited limited selectivityand efficacy. Here, we report that micheliolide (MCL), the activemetabolite of the clinical-stage candidate ACT001, inhibits STAT3through Cys367 (C367)-dependent covalent engagement within the DNA-bindingdomain (DBD)–a residue unique to STAT3 among STAT family members.This specific modification disrupts STAT3 phosphorylation at Tyr705,nuclear translocation, and transcriptional activity prior to any detectableinhibition of upstream JAK phosphorylation, a finding that suggeststhat STAT3 inhibition is not primarily driven by upstream kinase blockade.Guided by covalent docking, R-group enumeration, and relative bindingfree-energy calculations via free-energy perturbation (FEP), we havedesigned and synthesized a series of MCL analogs. These derivativesretain comparable antiproliferative potency but exhibit markedly improvedphysicochemical properties, including enhanced kinetic aqueous solubilityand optimized lipophilicity. Microscale thermophoresis confirms C367-dependentbinding, while microsecond-scale molecular dynamics simulations andpair-interaction energy analyses highlight stable complexes in whichelectrostatic interactions predominate, with major contributions fromkey residues (Lys365, Val366, and Ile368). This study establishesC367 as a druggable covalent site in the historically challengingSTAT3 DBD, validates FEP-guided optimization of covalent inhibitors,and identifies compound <strong>16</strong> as a promising next-generationlead with a superior drug-like profile for targeting STAT3-dependentcancers. All in all, these findings provide a foundation for advancingselective STAT3 DBD inhibitors toward clinical translation.</p>","PeriodicalId":44,"journal":{"name":"Journal of Chemical Information and Modeling ","volume":"66 19","pages":"12919–12934"},"PeriodicalIF":6.4,"publicationDate":"2026-10-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149049659","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 Multimodal Integration Framework for Taste and Odor Prediction and Flavor Molecule Discovery","authors":"Yingjie Song,&nbsp;Kun Tang,&nbsp;Juntao Wang,&nbsp;Guang Luo,&nbsp;Hanxiao Bao,&nbsp;Jingyuan Sun,&nbsp;Qilei Liu,&nbsp;Jian Du,&nbsp;Jing Hu,&nbsp;Lei Zhang","doi":"10.1021/acs.jcim.6c02181","DOIUrl":"10.1021/acs.jcim.6c02181","url":null,"abstract":"<p>Data-driven molecular property prediction is increasinglyusedto accelerate molecular discovery and computer-aided design. In flavorscience, accurate prediction of molecular taste and odor attributescould facilitate the screening and rational design of candidate flavormolecules. However, this task remains challenging because structure–flavorrelationships are complex, single molecular representations may provideincomplete property-relevant information, and the available labelsare highly imbalanced. To address these challenges, we implementedand evaluated Uni-Flavor, a multimodal prediction framework embeddedin an integrated computational discovery workflow. Uni-Flavor combinesthree-dimensional molecular representations extracted from a LoRA-fine-tunedUni-Mol2 model with stability-selected Mordred physicochemical descriptorsthrough a dual-branch fusion architecture incorporating modality-specificprojection, multiplicative late fusion, label-aware classification,and imbalance-aware optimization. Independent taste and odor modelswere trained separately under a common modeling and evaluation protocol,using curated datasets containing 17,202 taste molecules across sixcategories and 4,983 odor molecules across 138 labels. On held-outtest sets, the taste model achieved an accuracy of 95.63% and an AUROCof 97.35%, an AUPRC of 82.05%, and an F1 score of 76.32%, whereasthe corresponding values for the odor model were 96.03%, 89.97%, 38.00%and 36.94%. Branch-specific atom-level attribution analysis providedheuristic structural insights into the Uni-Mol2-derived componentof the predictions by associating attribution patterns with specificmolecular substructures. The trained models were subsequently usedto screen approximately 670,000 natural products and coupled witha ScaffoldCAMD strategy for the constrained design of synthetic flavorcandidates under chemical-validity, scaffold-assembly, and physicochemical-propertyconstraints, followed by preliminary structural safety filtering.Molecular docking was further used to examine plausible receptor–ligandinteractions and provide preliminary receptor-level support for selectedcandidates. This work presents a common multimodal modeling strategythat integrates three-dimensional molecular representations with physicochemicaldescriptors and is instantiated independently for multilabel tasteand odor prediction. Although developed for flavor-related prediction,evaluation on the SIDER dataset provides preliminary cross-task evidencethat the multimodal integration strategy can be applied to anotherimbalanced multilabel molecular property problem; broader transferabilityto other property domains remains to be established.</p>","PeriodicalId":44,"journal":{"name":"Journal of Chemical Information and Modeling ","volume":"66 19","pages":"12587–12603"},"PeriodicalIF":6.4,"publicationDate":"2026-10-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149090864","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":"Language of Toxicity: An eXplainable Artificial Intelligence Approach","authors":"Nicola Amoroso,&nbsp;Ester Pantaleo,&nbsp;Fulvio Ciriaco,&nbsp;Francesca Cutropia,&nbsp;Nicola Gambacorta,&nbsp;Fabrizio Mastrolorito,&nbsp;Alfonso Monaco,&nbsp;Angelica Orfino,&nbsp;Roberto Bellotti,&nbsp;Orazio Nicolotti","doi":"10.1021/acs.jcim.5c03214","DOIUrl":"10.1021/acs.jcim.5c03214","url":null,"abstract":"<p>Toxicity predictionin small molecules represents a fundamentalchallenge in drug development and chemical safety assessment. Traditionalapproaches heavily rely on predefined molecular descriptors or fingerprints,potentially limiting the ability to capture complex and nonlinearstructure–activity relationships. Here, we present a descriptor-free,language-inspired framework that can be applied to different toxicityprediction tasks within a unified architecture. The model proposedcombines a multiscale Convolutional Neural Network (CNN) layer tocapture chemical patterns at different scales and a Gated RecurrentUnit (GRU) layer to capture the sequential nature of these patterns.This architecture also exploits an attention mechanism that computesattention weights across the sequence, enabling the model to focuson the most relevant molecular substructures for toxicity prediction.Toxic and nontoxic chemicals, represented by canonical SMILES, areinvestigated as the words of two languages which have to be discriminated;using eight different end points, the model provided an accurate descriptionof toxicity patterns, with an average Area under the ROC curve (AUC)of 0.83 (min: 0.70, max: 0.94) under repeated cross-validation. Themodels were trained on relatively small data sets (∼1000 samples)and often strongly imbalanced, two important challenges that highlightthe opportunities for future improvement; moreover, the proposed attention-basedframework offers a representation of the molecular regions influencingmodel predictions, providing a basis for future investigations intotoxicity-related structural patterns and potentially supporting hypothesisgeneration in drug design or drug repurposing applications.</p>","PeriodicalId":44,"journal":{"name":"Journal of Chemical Information and Modeling ","volume":"66 19","pages":"12282–12290"},"PeriodicalIF":6.4,"publicationDate":"2026-10-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149090869","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":"Evaluation of the Complementarity of Three Major Databases of Protein–Ligand Binding Affinity Data","authors":"Haotian Gao,&nbsp;Fengfei Miao,&nbsp;Xinchong Chen,&nbsp;Jingyuan Li,&nbsp;Kexin Wu,&nbsp;Yanbei Li,&nbsp;Yan Li,&nbsp;Yifei Qi,&nbsp;Zhen Duan,&nbsp;Renxiao Wang","doi":"10.1021/acs.jcim.6c02172","DOIUrl":"10.1021/acs.jcim.6c02172","url":null,"abstract":"<p>Protein–ligand binding affinity data underpina varietyof tasks at the early stage of drug discovery. Public databases includingChEMBL, BindingDB, and PDBbind curate such data with distinct focuses,yet their data overlap and complementary attributes remain insufficientlyquantified. In this work, we systematically cross-mapped protein–ligandbinding data in PDBbind (version 2024) against data records sourcedfrom ChEMBL (version 35) and BindingDB (version 2025.08). Our analysisrevealed that merely 22.1% and 23.9% of PDBbind’s annotatedprotein–ligand complexes possess matching binding data recordsin ChEMBL and BindingDB, respectively, confirming that the PDBbinddataset is not a subset of the other two larger repositories. Thislimited overlap arises from fundamental differences in scope: ChEMBLand BindingDB center on drug discovery datasets, whereas the dataentries in PDBbind encompass a broader chemical and biological landscape.Our analysis further uncovered substantial data redundancy in ChEMBLand BindingDB, where 40%–50% of protein–ligand interactionpairs are associated with multiple, discrepant binding affinity values.Currently, no single resource achieves a comprehensive coverage ofavailable protein–ligand binding data, mandating researchersto adopt resource-selection strategies customized to their goals.While ChEMBL and BindingDB host extensive bioactivity archives, PDBbinddirectly links three-dimensional complex structures with carefullycurated binding affinity data. This unique value makes it best suitedfor structure-based studies.</p>","PeriodicalId":44,"journal":{"name":"Journal of Chemical Information and Modeling ","volume":"66 19","pages":"12604–12612"},"PeriodicalIF":6.4,"publicationDate":"2026-10-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149090893","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}