{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,4,28]],"date-time":"2026-04-28T15:13:04Z","timestamp":1777389184093,"version":"3.51.4"},"reference-count":36,"publisher":"Public Library of Science (PLoS)","issue":"3","license":[{"start":{"date-parts":[[2026,3,5]],"date-time":"2026-03-05T00:00:00Z","timestamp":1772668800000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"the general project of National Natural Science Foundation","award":["No. 62377006"],"award-info":[{"award-number":["No. 62377006"]}]},{"DOI":"10.13039\/501100017630","name":"Humanities and Social Sciences Youth Foundation, Ministry of Education","doi-asserted-by":"publisher","award":["No. 23YJA740033"],"award-info":[{"award-number":["No. 23YJA740033"]}],"id":[{"id":"10.13039\/501100017630","id-type":"DOI","asserted-by":"publisher"}]},{"name":"The key project of the Joint Fund for Free Exploration under Jilin Provincial Natural Science Foundation","award":["No. YDZJ202101ZYTS153"],"award-info":[{"award-number":["No. YDZJ202101ZYTS153"]}]}],"content-domain":{"domain":["www.plosone.org"],"crossmark-restriction":false},"short-container-title":["PLoS One"],"abstract":"<jats:p>Uveitis is a severe ocular inflammatory disease with complex immune\u2014mediated pathogenesis, posing significant challenges for drug discovery. While artificial intelligence has accelerated virtual screening, existing models often inadequately integrate heterogeneous molecular features or address disease\u2014specific mechanisms. To address these gaps, we propose MLGT (Multimodal Learning with Graph and molecular descriptors for Therapeutics), a novel graph attention network based on GATv2 that synergistically integrates molecular graph topology, bond attributes, and physicochemical descriptors within a unified deep learning framework. The model employs dynamic attention mechanisms to capture non\u2014local atomic interactions and a dual\u2014stream fusion module to combine graph embeddings with molecular descriptors. To mitigate data imbalance and overfitting, we implement label smoothing, class\u2014balanced sampling, and SMILES randomization. Evaluated on a rigorously curated Uveitis\u2014related compound dataset from ChEMBL, MLGT achieves state\u2014of\u2014the\u2014art performance: 97.7% accuracy, 97.2% F1 score, 96.1% recall, and an AUC\u2014ROC of 0.9156, surpassing existing graph learning and classical machine learning benchmarks. Ablation studies confirm the essential roles of multimodal fusion and attention mechanisms. This study provides an efficient, attention\u2014based computational tool for targeted Uveitis drug screening and establishes a scalable AI\u2014driven paradigm for precision drug discovery in complex diseases.<\/jats:p>","DOI":"10.1371\/journal.pone.0343159","type":"journal-article","created":{"date-parts":[[2026,3,5]],"date-time":"2026-03-05T18:23:55Z","timestamp":1772735035000},"page":"e0343159","update-policy":"https:\/\/doi.org\/10.1371\/journal.pone.corrections_policy","source":"Crossref","is-referenced-by-count":1,"title":["MLGT: A multimodal graph attention network for virtual screening of anti\u2014Uveitis drugs"],"prefix":"10.1371","volume":"21","author":[{"given":"Yu","family":"Sun","sequence":"first","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0009-0007-3537-6787","authenticated-orcid":true,"given":"Yihang","family":"Qin","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Wenhao","family":"Chen","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"340","published-online":{"date-parts":[[2026,3,5]]},"reference":[{"issue":"7","key":"pone.0343159.ref001","doi-asserted-by":"crossref","first-page":"3011","DOI":"10.4049\/jimmunol.1402650","article-title":"Retina-specific T regulatory cells bring about resolution and maintain remission of autoimmune uveitis","volume":"194","author":"PB Silver","year":"2015","journal-title":"J Immunol"},{"issue":"4","key":"pone.0343159.ref002","doi-asserted-by":"crossref","first-page":"339","DOI":"10.3109\/09273948.2014.981548","article-title":"Causes of uveitis in a tertiary center in chile: A cross-sectional retrospective review","volume":"23","author":"P Liberman","year":"2015","journal-title":"Ocul Immunol Inflamm"},{"issue":"4","key":"pone.0343159.ref003","doi-asserted-by":"crossref","first-page":"669","DOI":"10.1016\/j.survophthal.2023.02.007","article-title":"Artificial intelligence in uveitis: A comprehensive review","volume":"68","author":"LF Nakayama","year":"2023","journal-title":"Surv Ophthalmol"},{"issue":"11","key":"pone.0343159.ref004","doi-asserted-by":"crossref","first-page":"3746","DOI":"10.3390\/jcm12113746","article-title":"Diagnosis, classification, and assessment of the underlying etiology of Uveitis by Artificial Intelligence: A systematic review","volume":"12","author":"R Jacquot","year":"2023","journal-title":"J Clin Med"},{"issue":"5","key":"pone.0343159.ref005","doi-asserted-by":"crossref","first-page":"992","DOI":"10.1038\/s41433-024-03530-2","article-title":"Neural networks for predicting etiological diagnosis of uveitis","volume":"39","author":"R Jacquot","year":"2025","journal-title":"Eye (Lond)"},{"key":"pone.0343159.ref006","doi-asserted-by":"crossref","first-page":"77","DOI":"10.1016\/j.ajo.2018.02.019","article-title":"Autoimmunity, autoinflammation, and infection in uveitis","volume":"189","author":"JV Forrester","year":"2018","journal-title":"American Journal of Ophthalmology"},{"issue":"10","key":"pone.0343159.ref007","doi-asserted-by":"crossref","first-page":"3931","DOI":"10.1167\/iovs.17-21493","article-title":"Uvemaster: A mobile app-based decision support system for the differential diagnosis of Uveitis","volume":"58","author":"JA Gegundez-Fernandez","year":"2017","journal-title":"Invest Ophthalmol Vis Sci"},{"issue":"1","key":"pone.0343159.ref008","doi-asserted-by":"crossref","DOI":"10.1038\/s43586-024-00294-7","article-title":"Graph neural networks","volume":"4","author":"G Corso","year":"2024","journal-title":"Nat Rev Methods Primers"},{"issue":"12","key":"pone.0343159.ref009","doi-asserted-by":"crossref","first-page":"1511","DOI":"10.3390\/genes15121511","article-title":"GATv2EPI: Predicting enhancer-promoter interactions with a dynamic graph attention network","volume":"15","author":"T Zhang","year":"2024","journal-title":"Genes (Basel)"},{"key":"pone.0343159.ref010","first-page":"162","article-title":"Distribution-balanced loss for multi-label classification in long-tailed datasets","author":"T Wu","year":"2020","journal-title":"In: European conference on computer vision"},{"issue":"1","key":"pone.0343159.ref011","doi-asserted-by":"crossref","first-page":"101","DOI":"10.1186\/s40537-021-00492-0","article-title":"Text data augmentation for deep learning","volume":"8","author":"C Shorten","year":"2021","journal-title":"Journal of Big Data"},{"key":"pone.0343159.ref012","first-page":"119","article-title":"Understanding and bridging the gaps in current GNN performance optimizations","author":"K Huang","year":"2021","journal-title":"In: Proceedings of the 26th ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming"},{"key":"pone.0343159.ref013","doi-asserted-by":"crossref","first-page":"1427144","DOI":"10.3389\/fmed.2025.1427144","article-title":"Natural active ingredients targeted inflammatory cytokines and major blinding eye diseases: a two-sample Mendelian randomization and molecular docking analysis","volume":"12","author":"Y Shao","year":"2025","journal-title":"Front Med (Lausanne)"},{"issue":"14","key":"pone.0343159.ref014","doi-asserted-by":"crossref","DOI":"10.1103\/PhysRevB.95.144110","article-title":"Representation of compounds for machine-learning prediction of physical properties","volume":"95","author":"A Seko","year":"2017","journal-title":"Phys Rev B"},{"issue":"6","key":"pone.0343159.ref015","doi-asserted-by":"crossref","first-page":"2680","DOI":"10.3390\/ijms26062680","article-title":"Enhancing HCV NS3 inhibitor classification with optimized molecular fingerprints using random forest","volume":"26","author":"S Atasever","year":"2025","journal-title":"Int J Mol Sci"},{"key":"pone.0343159.ref016","first-page":"10654","article-title":"Universal graph convolutional networks","volume":"34","author":"Z Yu","year":"2021","journal-title":"Advances in neural information processing systems"},{"key":"pone.0343159.ref017","article-title":"Graph attention networks","author":"P Veli\u010dkovi\u0107","year":"2017","journal-title":"arxiv preprint"},{"issue":"1","key":"pone.0343159.ref018","doi-asserted-by":"crossref","first-page":"562","DOI":"10.1109\/JBHI.2022.3219213","article-title":"Predicting drug-target interactions over heterogeneous information network","volume":"27","author":"X Su","year":"2023","journal-title":"IEEE J Biomed Health Inform"},{"issue":"1","key":"pone.0343159.ref019","doi-asserted-by":"crossref","first-page":"573","DOI":"10.1109\/JBHI.2022.3217433","article-title":"PPAEDTI: Personalized propagation auto-encoder model for predicting drug-target interactions","volume":"27","author":"Y-C Li","year":"2023","journal-title":"IEEE J Biomed Health Inform"},{"issue":"7","key":"pone.0343159.ref020","doi-asserted-by":"crossref","first-page":"4281","DOI":"10.1109\/JBHI.2024.3383591","article-title":"Motif-Aware miRNA-Disease association prediction via hierarchical attention network","volume":"28","author":"B-W Zhao","year":"2024","journal-title":"IEEE J Biomed Health Inform"},{"key":"pone.0343159.ref021","doi-asserted-by":"crossref","first-page":"2924","DOI":"10.1016\/j.csbj.2024.06.032","article-title":"A heterogeneous information network learning model with neighborhood-level structural representation for predicting lncRNA-miRNA interactions","volume":"23","author":"B-W Zhao","year":"2024","journal-title":"Comput Struct Biotechnol J"},{"key":"pone.0343159.ref022","article-title":"LLM-DDI: leveraging large language models for drug-drug interaction prediction on biomedical knowledge graph","author":"D Li","year":"2025","journal-title":"IEEE Journal of Biomedical and Health Informatics"},{"key":"pone.0343159.ref023","article-title":"Multi-view contrastive learning for drug-drug interaction event prediction","author":"D Li","year":"2025","journal-title":"IEEE J Biomed Health Inform"},{"issue":"1","key":"pone.0343159.ref024","doi-asserted-by":"crossref","first-page":"7","DOI":"10.1186\/s12967-016-1103-6","article-title":"Prediction of anti-inflammatory proteins\/peptides: an insilico approach","volume":"15","author":"S Gupta","year":"2017","journal-title":"J Transl Med"},{"issue":"12","key":"pone.0343159.ref025","doi-asserted-by":"crossref","first-page":"103373","DOI":"10.1016\/j.drudis.2022.103373","article-title":"Deep learning methods for molecular representation and property prediction","volume":"27","author":"Z Li","year":"2022","journal-title":"Drug Discov Today"},{"key":"pone.0343159.ref026","doi-asserted-by":"crossref","DOI":"10.1093\/nar\/gkr777","article-title":"ChEMBL: a large-scale bioactivity database for drug discovery","volume":"40","author":"A Gaulton","year":"2012","journal-title":"Nucleic Acids Res"},{"issue":"6","key":"pone.0343159.ref027","doi-asserted-by":"crossref","first-page":"1048","DOI":"10.4218\/etrij.13.2013.0081","article-title":"An immersive augmented\u2010reality\u2010based e\u2010learning system based on dynamic threshold marker method","volume":"35","author":"S Lim","year":"2013","journal-title":"ETRI Journal"},{"issue":"1","key":"pone.0343159.ref028","doi-asserted-by":"crossref","first-page":"77","DOI":"10.1021\/acs.jcim.9b00727","article-title":"DeepScaffold: A comprehensive tool for scaffold-based de novo drug discovery using deep learning","volume":"60","author":"Y Li","year":"2020","journal-title":"J Chem Inf Model"},{"issue":"4","key":"pone.0343159.ref029","doi-asserted-by":"crossref","first-page":"733","DOI":"10.1111\/1756-2171.12036","article-title":"Targeted search and the long tail effect","volume":"44","author":"H Yang","year":"2013","journal-title":"The RAND J of Economics"},{"issue":"22","key":"pone.0343159.ref030","doi-asserted-by":"crossref","first-page":"224106","DOI":"10.1063\/1.4984965","article-title":"The entropy of a complex molecule","volume":"146","author":"G Faure","year":"2017","journal-title":"J Chem Phys"},{"issue":"3","key":"pone.0343159.ref031","first-page":"307","article-title":"Receiver operating characteristic (ROC) methodology: the state of the art","volume":"29","author":"JA Hanley","year":"1989","journal-title":"Crit Rev Diagn Imaging"},{"key":"pone.0343159.ref032","first-page":"507","volume-title":"Drug\u2014Target Interaction Prediction Based on Attentive FP and Word2vec [C]\/\/International Conference on Intelligent Computing","author":"Y Lei","year":"2022"},{"issue":"1","key":"pone.0343159.ref033","doi-asserted-by":"crossref","first-page":"9","DOI":"10.1021\/acs.jcim.3c01250","article-title":"Chemprop: A machine learning package for chemical property prediction","volume":"64","author":"E Heid","year":"2024","journal-title":"J Chem Inf Model"},{"issue":"3","key":"pone.0343159.ref034","doi-asserted-by":"crossref","first-page":"371","DOI":"10.1038\/s41551-024-01312-5","article-title":"Interpretable identification of cancer genes across biological networks via transformer-powered graph representation learning","volume":"9","author":"X Su","year":"2025","journal-title":"Nat Biomed Eng"},{"issue":"8","key":"pone.0343159.ref035","doi-asserted-by":"crossref","first-page":"5730","DOI":"10.1109\/TSMC.2025.3572738","article-title":"Link-based attributed graph clustering via approximate generative bayesian learning","volume":"55","author":"Y Yang","year":"2025","journal-title":"IEEE Trans Syst Man Cybern, Syst"},{"issue":"13","key":"pone.0343159.ref036","doi-asserted-by":"crossref","first-page":"7068","DOI":"10.3390\/app15137068","article-title":"Context-Aware multimodal fusion with sensor-augmented cross-modal learning: The BLAF architecture for robust chinese homophone disambiguation in dynamic environments","volume":"15","author":"Y Sun","year":"2025","journal-title":"Applied Sciences"}],"container-title":["PLOS One"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pone.0343159","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,3,5]],"date-time":"2026-03-05T18:24:01Z","timestamp":1772735041000},"score":1,"resource":{"primary":{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pone.0343159"}},"subtitle":[],"editor":[{"given":"Kayode","family":"Raheem","sequence":"first","affiliation":[],"role":[{"role":"editor","vocabulary":"crossref"}]}],"short-title":[],"issued":{"date-parts":[[2026,3,5]]},"references-count":36,"journal-issue":{"issue":"3","published-online":{"date-parts":[[2026,3,5]]}},"URL":"https:\/\/doi.org\/10.1371\/journal.pone.0343159","relation":{},"ISSN":["1932-6203"],"issn-type":[{"value":"1932-6203","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,3,5]]}}}