{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,14]],"date-time":"2026-10-14T18:44:29Z","timestamp":1792003469120,"version":"4.4.0"},"reference-count":32,"publisher":"Public Library of Science (PLoS)","issue":"10","license":[{"start":{"date-parts":[[2026,10,14]],"date-time":"2026-10-14T00:00:00Z","timestamp":1791936000000},"content-version":"am","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["62302198"],"award-info":[{"award-number":["62302198"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"name":"the Funded by Basic Research Program of Jiangsu","award":["BK20231035"],"award-info":[{"award-number":["BK20231035"]}]},{"DOI":"10.13039\/501100012226","name":"Fundamental Research Funds for the Central Universities","doi-asserted-by":"publisher","award":["JUSRP124014"],"award-info":[{"award-number":["JUSRP124014"]}],"id":[{"id":"10.13039\/501100012226","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["www.ploscompbiol.org"],"crossmark-restriction":false},"short-container-title":["PLoS Comput Biol"],"abstract":"<jats:p>\n                    Drug-drug interactions (DDIs) are critical in pharmacology, as synergistic or antagonistic effects between co-administered drugs can enhance therapeutic outcomes or cause toxicity, posing major challenges for drug safety assessment and new drug development. Current molecular representation methods often fail to capture dynamic substructure relationships or semantic interactions among functional groups. To overcome these limitations, this study proposes MEGH-DDI, a multimodal framework that integrates complementary molecular representation learning with graph-structure dynamic perception. The framework employs a dual-channel semantic module that combines Mol2vec-based substructure representations with MolT5 chemical language embeddings, while SHAP-based dynamic feature selection reduces redundancy. For fine-grained structural modeling, we introduce an edge-aware GCNII equipped with bond-level attention to distinguish different types of chemical bonds. Furthermore, the Hop2Token module converts multi-hop neighborhoods into token sequences, enabling Transformer-based modeling of high-order substructure dependencies. Experimental results demonstrate that MEGH-DDI achieves competitive performance across both transductive and inductive settings. On the DrugBank dataset, MEGH-DDI attains an accuracy of 98.36% in the transductive task, outperforming earlier baselines and surpassing recent state-of-the-art methods, with both AUC and AP exceeding 99.7%. On the ChChMiner dataset, MEGH-DDI achieves an F1 score of 99.05%, improving over the best existing method by 2.74 percentage points. In the inductive settings, MEGH-DDI achieves competitive performance across most evaluation metrics on both datasets, showing advantages over competing methods on several key metrics. These results demonstrate the effectiveness of MEGH-DDI in predicting DDI involving previously unseen drugs and its potential to generalize to complex DDI prediction scenarios. To enhance interpretability, MEGH-DDI incorporates a three-level mechanism spanning atoms, chemical bonds, and topological paths, highlighting key atoms and bond contributions. Overall, MEGH-DDI provides an efficient, accurate, and interpretable solution for drug safety assessment and new drug discovery. Code and datasets are available at\n                    <jats:ext-link xmlns:xlink=\"http:\/\/www.w3.org\/1999\/xlink\" ext-link-type=\"uri\" xlink:href=\"https:\/\/github.com\/Lilyjoys\/MEGH-DDI\" xlink:type=\"simple\">https:\/\/github.com\/Lilyjoys\/MEGH-DDI<\/jats:ext-link>\n                    , and a free web platform is provided at\n                    <jats:ext-link xmlns:xlink=\"http:\/\/www.w3.org\/1999\/xlink\" ext-link-type=\"uri\" xlink:href=\"http:\/\/www.edgh-ddi.com\" xlink:type=\"simple\">http:\/\/www.edgh-ddi.com<\/jats:ext-link>\n                    .\n                  <\/jats:p>","DOI":"10.1371\/journal.pcbi.1014393","type":"journal-article","created":{"date-parts":[[2026,10,14]],"date-time":"2026-10-14T17:45:52Z","timestamp":1791999952000},"page":"e1014393","update-policy":"https:\/\/doi.org\/10.1371\/journal.pcbi.corrections_policy","source":"Crossref","is-referenced-by-count":0,"title":["MEGH-DDI: MolT5-driven edge-aware graph learning with Hop2Token representation for drug\u2013drug interaction prediction"],"prefix":"10.1371","volume":"22","author":[{"given":"Li","family":"Zhou","sequence":"first","affiliation":[{"name":"School of Artificial Intelligence and Computer Science, Jiangnan University","place":["Wuxi, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0003-4021-3836","authenticated-orcid":true,"given":"Yun","family":"Zuo","sequence":"additional","affiliation":[{"name":"School of Artificial Intelligence and Computer Science, Jiangnan University","place":["Wuxi, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Wenying","family":"He","sequence":"additional","affiliation":[{"name":"School of Artificial Intelligence, Hebei University of Technology","place":["Tianjin, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Zhaohong","family":"Deng","sequence":"additional","affiliation":[{"name":"School of Artificial Intelligence and Computer Science, Jiangnan University","place":["Wuxi, China"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"340","published-online":{"date-parts":[[2026,10,14]]},"reference":[{"key":"pcbi.1014393.ref001","doi-asserted-by":"crossref","first-page":"107996","DOI":"10.1016\/j.cnsns.2024.107996","article-title":"Mathematical modeling and dynamic analysis for cancer resistance incorporating persister cells","volume":"134","author":"K Qi","year":"2024","journal-title":"Commun Nonlinear Sci Num Simul"},{"issue":"19","key":"pcbi.1014393.ref002","doi-asserted-by":"crossref","first-page":"16532","DOI":"10.1021\/acsomega.3c00222","article-title":"Single inhibitors versus dual inhibitors: Role of HDAC in cancer","volume":"8","author":"R Roy","year":"2023","journal-title":"ACS Omega"},{"issue":"2","key":"pcbi.1014393.ref003","doi-asserted-by":"crossref","first-page":"104","DOI":"10.1038\/s41565-022-01265-3","article-title":"Multidrug nanomedicine","volume":"18","author":"AM Sofias","year":"2023","journal-title":"Nat Nanotechnol"},{"issue":"12","key":"pcbi.1014393.ref004","article-title":"Response by Ward et al to letter regarding article, \u201cStatin Toxicity: Mechanistic insights and clinical implications\u201d","volume":"124","author":"NC Ward","year":"2019","journal-title":"Circ Res"},{"issue":"4","key":"pcbi.1014393.ref005","article-title":"Comprehensive evaluation of deep and graph learning on drug-drug interactions prediction","volume":"24","author":"X Lin","year":"2023","journal-title":"Brief Bioinform"},{"key":"pcbi.1014393.ref006","doi-asserted-by":"crossref","first-page":"1608657","DOI":"10.3389\/fphar.2025.1608657","article-title":"A retrospective observational study on case reports of adverse drug reactions (ADRs) to tirzepatide","volume":"16","author":"M Huang","year":"2025","journal-title":"Front Pharmacol"},{"key":"pcbi.1014393.ref007","doi-asserted-by":"crossref","first-page":"592","DOI":"10.1038\/msb.2012.26","article-title":"INDI: A computational framework for inferring drug interactions and their associated recommendations","volume":"8","author":"A Gottlieb","year":"2012","journal-title":"Mol Syst Biol"},{"key":"pcbi.1014393.ref008","doi-asserted-by":"crossref","DOI":"10.1136\/amiajnl-2013-002512","article-title":"Machine learning-based prediction of drug-drug interactions by integrating drug phenotypic, therapeutic, chemical, and genomic properties","volume":"21","author":"F Cheng","year":"2014","journal-title":"J Am Med Inform Assoc"},{"issue":"3","key":"pcbi.1014393.ref009","doi-asserted-by":"crossref","first-page":"823","DOI":"10.1109\/TCBB.2020.2979959","article-title":"Biomedical knowledge graphs construction from conditional statements","volume":"18","author":"T Jiang","year":"2021","journal-title":"IEEE\/ACM Trans Comput Biol Bioinform"},{"key":"pcbi.1014393.ref010","doi-asserted-by":"crossref","DOI":"10.1093\/nar\/gkx1037","article-title":"DrugBank 5.0: A major update to the DrugBank database for 2018","volume":"46","author":"DS Wishart","year":"2018","journal-title":"Nucleic Acids Res"},{"issue":"4","key":"pcbi.1014393.ref011","doi-asserted-by":"crossref","DOI":"10.1002\/wsbm.1417","article-title":"PharmGKB: A worldwide resource for pharmacogenomic information","volume":"10","author":"JM Barbarino","year":"2018","journal-title":"Wiley Interdiscip Rev Syst Biol Med"},{"key":"pcbi.1014393.ref012","doi-asserted-by":"crossref","unstructured":"Karim MdR, Cochez M, Jares JB, Uddin M, Beyan O, Decker S. Drug-drug interaction prediction based on knowledge graph embeddings and convolutional-LSTM network. In: Proceedings of the 10th ACM International Conference on bioinformatics, computational biology and health informatics. 2019. 113\u201323. https:\/\/doi.org\/10.1145\/3307339.3342161","DOI":"10.1145\/3307339.3342161"},{"issue":"18","key":"pcbi.1014393.ref013","doi-asserted-by":"crossref","first-page":"2988","DOI":"10.1093\/bioinformatics\/btab207","article-title":"SumGNN: Multi-typed drug interaction prediction via efficient knowledge graph summarization","volume":"37","author":"Y Yu","year":"2021","journal-title":"Bioinformatics"},{"issue":"7","key":"pcbi.1014393.ref014","doi-asserted-by":"crossref","first-page":"3370","DOI":"10.1021\/acs.jcim.5c00275","article-title":"Application of large language models in drug-induced osteotoxicity prediction","volume":"65","author":"Y-Q Chen","year":"2025","journal-title":"J Chem Inf Model"},{"issue":"6","key":"pcbi.1014393.ref015","doi-asserted-by":"crossref","DOI":"10.1093\/bib\/bbab133","article-title":"SSI-DDI: Substructure-substructure interactions for drug-drug interaction prediction","volume":"22","author":"AK Nyamabo","year":"2021","journal-title":"Brief Bioinform"},{"issue":"29","key":"pcbi.1014393.ref016","doi-asserted-by":"crossref","first-page":"8693","DOI":"10.1039\/D2SC02023H","article-title":"Learning size-adaptive molecular substructures for explainable drug-drug interaction prediction by substructure-aware graph neural network","volume":"13","author":"Z Yang","year":"2022","journal-title":"Chem Sci"},{"issue":"1","key":"pcbi.1014393.ref017","article-title":"MDF-SA-DDI: Predicting drug-drug interaction events based on multi-source drug fusion, multi-source feature fusion and transformer self-attention mechanism","volume":"23","author":"S Lin","year":"2022","journal-title":"Brief Bioinform"},{"issue":"1","key":"pcbi.1014393.ref018","doi-asserted-by":"crossref","DOI":"10.1093\/bib\/bbab441","article-title":"Drug-drug interaction prediction with learnable size-adaptive molecular substructures","volume":"23","author":"AK Nyamabo","year":"2022","journal-title":"Brief Bioinform"},{"issue":"1","key":"pcbi.1014393.ref019","article-title":"DSN-DDI: an accurate and generalized framework for drug-drug interaction prediction by dual-view representation learning","volume":"24","author":"Z Li","year":"2023","journal-title":"Brief Bioinform"},{"key":"pcbi.1014393.ref020","doi-asserted-by":"crossref","first-page":"102640","DOI":"10.1016\/j.artmed.2023.102640","article-title":"DDI-GCN: Drug-drug interaction prediction via explainable graph convolutional networks","volume":"144","author":"Y Zhong","year":"2023","journal-title":"Artif Intell Med"},{"issue":"1","key":"pcbi.1014393.ref021","article-title":"A dual graph neural network for drug-drug interactions prediction based on molecular structure and interactions","volume":"19","author":"M Ma","year":"2023","journal-title":"PLoS Comput Biol"},{"issue":"1","key":"pcbi.1014393.ref022","doi-asserted-by":"crossref","first-page":"569","DOI":"10.1109\/JBHI.2023.3335402","article-title":"PEB-DDI: A task-specific dual-view substructural learning framework for drug-drug interaction prediction","volume":"28","author":"X Shen","year":"2024","journal-title":"IEEE J Biomed Health Inform"},{"key":"pcbi.1014393.ref023","article-title":"Stanford biomedical network dataset collection","author":"M Zitnik","year":"2018","journal-title":"Biosn Datasets Stanford Biomed Netw Dataset Collect"},{"issue":"1","key":"pcbi.1014393.ref024","doi-asserted-by":"crossref","first-page":"51","DOI":"10.1186\/s13321-020-00456-1","article-title":"An open source chemical structure curation pipeline using RDKit","volume":"12","author":"AP Bento","year":"2020","journal-title":"J Cheminform"},{"issue":"1","key":"pcbi.1014393.ref025","doi-asserted-by":"crossref","first-page":"27","DOI":"10.1021\/acs.jcim.7b00616","article-title":"Mol2vec: Unsupervised machine learning approach with chemical intuition","volume":"58","author":"S Jaeger","year":"2018","journal-title":"J Chem Inf Model"},{"issue":"1","key":"pcbi.1014393.ref026","doi-asserted-by":"crossref","first-page":"155","DOI":"10.1017\/S1351324916000334","article-title":"Word2Vec","volume":"23","author":"KW Church","year":"2017","journal-title":"Natural Lang Eng"},{"key":"pcbi.1014393.ref027","doi-asserted-by":"crossref","unstructured":"Edwards C, Lai T, Ros K, Honke G, Cho K, Ji H. Translation between molecules and natural language. 2022. https:\/\/arxiv.org\/abs\/2204.11817","DOI":"10.18653\/v1\/2022.emnlp-main.26"},{"key":"pcbi.1014393.ref028","unstructured":"Chen M, Wei Z, Huang Z, Ding B, Li Y. Simple and deep graph convolutional networks. 2020."},{"key":"pcbi.1014393.ref029","author":"J Chen","year":"2022"},{"key":"pcbi.1014393.ref030","article-title":"Hierarchical graph representation learning with differentiable pooling","volume":"31","author":"Z Ying","year":"2018","journal-title":"Adv Neural Inform Process Syst"},{"key":"pcbi.1014393.ref031","unstructured":"Lee J, Lee I, Kang J. Self-attention graph pooling. In: Proceedings of the 36th International conference on machine learning (ICML), 2019. 3734\u201343."},{"key":"pcbi.1014393.ref032","doi-asserted-by":"crossref","unstructured":"Wang Y, Min Y, Chen X, Wu J. Multi-view graph contrastive representation learning for drug-drug interaction prediction. In: Proceedings of the web conference 2021. 2021. 2921\u201333. https:\/\/doi.org\/10.1145\/3442381.3449786","DOI":"10.1145\/3442381.3449786"}],"container-title":["PLOS Computational Biology"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pcbi.1014393","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,10,14]],"date-time":"2026-10-14T17:45:59Z","timestamp":1791999959000},"score":1,"resource":{"primary":{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pcbi.1014393"}},"subtitle":[],"editor":[{"given":"Amr Ali","family":"Alhossary","sequence":"first","affiliation":[{"name":"Wesleyan University","place":["UNITED STATES OF AMERICA"]}],"role":[{"vocabulary":"crossref","role":"editor"}]}],"short-title":[],"issued":{"date-parts":[[2026,10,14]]},"references-count":32,"journal-issue":{"issue":"10","published-online":{"date-parts":[[2026,10,14]]}},"URL":"https:\/\/doi.org\/10.1371\/journal.pcbi.1014393","relation":{},"ISSN":["1553-7358"],"issn-type":[{"value":"1553-7358","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,10,14]]}}}