{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,12]],"date-time":"2026-10-12T13:45:50Z","timestamp":1791812750457,"version":"4.3.4"},"reference-count":36,"publisher":"American Chemical Society (ACS)","issue":"19","license":[{"start":{"date-parts":[[2026,9,30]],"date-time":"2026-09-30T00:00:00Z","timestamp":1790726400000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2026,9,30]],"date-time":"2026-09-30T00:00:00Z","timestamp":1790726400000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"},{"start":{"date-parts":[[2026,9,30]],"date-time":"2026-09-30T00:00:00Z","timestamp":1790726400000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-045"}],"funder":[{"DOI":"10.13039\/501100012166","name":"National Key Research and Development Program of China","doi-asserted-by":"publisher","award":["2025YFA0922300"],"award-info":[{"award-number":["2025YFA0922300"]}],"id":[{"id":"10.13039\/501100012166","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2026,10,12]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:p>Enzymes play a central role in green chemistry and biomanufacturing. However, precise enzyme selection via Enzyme Commission (EC) number prediction remains a critical bottleneck between the retrosynthetic pathway planning and experimental implementation. Existing computational methods predominantly rely on single-modality representations that either lose spatial topological structures or treat reactions as static snapshots, failing to capture the explicit atom-level state transitions in chemical bond evolution. To address these limitations, we propose GMSF, a dual-path multimodal deep learning framework. GMSF features two core innovations: Difference Graph Encoding, which leverages atom\u2013atom mapping (AAM) to compute node feature differences and extract structural difference signatures of reaction centers; and Multiscale Sequence Encoding, which uses AAM-aligned node mappings to hierarchically capture chemical semantics. Experimental results on the ECREACT dataset demonstrate that GMSF achieves an accuracy of 93.17% in the highly challenging level 3 (subsubclass) enzyme function prediction, outperforming the state-of-the-art (SOTA) baseline by a margin of 1.36%. Furthermore, its successful application to real-world metabolic networks in the BioCyc database validates its immense potential for deciphering complex biochemical reactions, serving as a powerful computational screening tool to narrow down catalytic mechanisms and coenzyme dependencies prior to the exact enzyme selection.<\/jats:p>","DOI":"10.1021\/acs.jcim.6c02672","type":"journal-article","created":{"date-parts":[[2026,9,30]],"date-time":"2026-09-30T16:03:49Z","timestamp":1790784229000},"page":"12904-12918","source":"Crossref","is-referenced-by-count":0,"title":["GMSF: A Dual-Path\nMultimodal Framework for Enzyme\nFunction Prediction via Difference Graph Encoding and Multiscale Semantic\nFusion"],"prefix":"10.1021","volume":"66","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-1621-2337","authenticated-orcid":true,"given":"Xin","family":"Zhao","sequence":"first","affiliation":[{"name":"Tianjin University , , 92 Weijin Road , Nankai District, ,","place":["Tianjin, China, 300072"]},{"name":"Tianjin University , , 92 Weijin Road , Nankai District, ,","place":["Tianjin, China, 300072"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Haoshu","family":"Chen","sequence":"additional","affiliation":[{"name":"Tianjin University , , 92 Weijin Road , Nankai District, ,","place":["Tianjin, China, 300072"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-2317-644X","authenticated-orcid":true,"given":"Tao","family":"Zhang","sequence":"additional","affiliation":[{"name":"Tianjin University , , 92 Weijin Road , Nankai District, ,","place":["Tianjin, China, 300072"]},{"name":"Tianjin University , , 92 Weijin Road , Nankai District, ,","place":["Tianjin, China, 300072"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yahui","family":"Cao","sequence":"additional","affiliation":[{"name":"Tianjin University , , 92 Weijin Road , Nankai District, ,","place":["Tianjin, China, 300072"]},{"name":"Tianjin University , , 92 Weijin Road , Nankai District, ,","place":["Tianjin, China, 300072"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Haotong","family":"Li","sequence":"additional","affiliation":[{"name":"Tianjin University , , 92 Weijin Road , Nankai District, ,","place":["Tianjin, China, 300072"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Zhuoran","family":"Song","sequence":"additional","affiliation":[{"name":"Tianjin University , , 92 Weijin Road , Nankai District, ,","place":["Tianjin, China, 300072"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-4121-3048","authenticated-orcid":true,"given":"Bingzhi","family":"Li","sequence":"additional","affiliation":[{"name":"Tianjin University , , 92 Weijin Road , Nankai District, ,","place":["Tianjin, China, 300072"]},{"name":"Tianjin University , , 92 Weijin Road , Nankai District, ,","place":["Tianjin, China, 300072"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Shuo","family":"Zheng","sequence":"additional","affiliation":[{"name":"Wuhan University , , 299 Bayi Road , Wuchang District, , ,","place":["Wuhan, Hubei, China, 430072"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"316","published-online":{"date-parts":[[2026,9,30]]},"reference":[{"issue":"2","key":"2026101208080962500_cit1","doi-asserted-by":"publisher","first-page":"801","DOI":"10.1021\/acs.chemrev.7b00203","article-title":"Role of biocatalysis in sustainable chemistry","volume":"118","author":"Sheldon","year":"2018","journal-title":"Chem. Rev"},{"issue":"7698","key":"2026101208080962500_cit2","doi-asserted-by":"publisher","first-page":"604","DOI":"10.1038\/nature25978","article-title":"Planning chemical\nsyntheses with\ndeep neural networks and symbolic AI","volume":"555","author":"Segler","year":"2018","journal-title":"Nature"},{"issue":"6639","key":"2026101208080962500_cit3","doi-asserted-by":"publisher","first-page":"1358","DOI":"10.1126\/science.adf2465","article-title":"Enzyme function prediction using contrastive\nlearning","volume":"379","author":"Yu","year":"2023","journal-title":"Science"},{"key":"2026101208080962500_cit4","article-title":"ChemBERTa:\nlarge-scale\nself-supervised pretraining for molecular property prediction","author":"Chithrananda","year":"2020","journal-title":"arXiv"},{"key":"2026101208080962500_cit5","doi-asserted-by":"crossref","first-page":"429","DOI":"10.1145\/3307339.3342186","article-title":"Smiles-bert: large scale unsupervised pre-training\nfor molecular property prediction","volume-title":"Proceedings\nof the 10th ACM international conference on bioinformatics, computational\nbiology and health informatics","author":"Wang","year":"2019"},{"issue":"1","key":"2026101208080962500_cit6","doi-asserted-by":"publisher","first-page":"56","DOI":"10.1186\/s13321-020-00460-5","article-title":"Molecular\nrepresentations\nin AI-driven drug discovery: a review and practical guide","volume":"12","author":"David","year":"2020","journal-title":"J. Cheminf"},{"issue":"20","key":"2026101208080962500_cit7","doi-asserted-by":"publisher","DOI":"10.1016\/j.heliyon.2024.e39038","article-title":"Transformer-based\nmodels for chemical SMILES representation:\nA comprehensive literature review","volume":"10","author":"Mswahili","year":"2024","journal-title":"Heliyon"},{"key":"2026101208080962500_cit8","article-title":"From words to molecules: A survey of\nlarge language models in chemistry","author":"Liao","year":"2024","journal-title":"arXiv"},{"issue":"1","key":"2026101208080962500_cit9","doi-asserted-by":"publisher","first-page":"38","DOI":"10.1186\/s13321-024-00827-y","article-title":"A general model for\npredicting enzyme functions\nbased on enzymatic reactions","volume":"16","author":"Qian","year":"2024","journal-title":"J. Cheminf"},{"issue":"1","key":"2026101208080962500_cit10","doi-asserted-by":"publisher","first-page":"2","DOI":"10.1186\/s13321-024-00944-8","article-title":"CLAIRE: a contrastive\nlearning-based predictor for EC number of chemical reactions","volume":"17","author":"Zeng","year":"2025","journal-title":"J. Cheminf"},{"issue":"12","key":"2026101208080962500_cit11","doi-asserted-by":"publisher","first-page":"3588","DOI":"10.1039\/D5DD00187K","article-title":"Leveraging large language models\nfor enzymatic reaction prediction and characterization","volume":"4","author":"Di Fruscia","year":"2025","journal-title":"Digital Discovery"},{"issue":"9","key":"2026101208080962500_cit12","doi-asserted-by":"publisher","first-page":"937","DOI":"10.1080\/17460441.2021.1915982","article-title":"Critical assessment\nof AI in drug discovery","volume":"16","author":"Walters","year":"2021","journal-title":"Expert Opin. Drug Discovery"},{"key":"2026101208080962500_cit13","article-title":"Predicting organic\nreaction outcomes with weisfeiler-lehman network","volume-title":"Advances in neural information processing systems 30","author":"Jin","year":"2017"},{"key":"2026101208080962500_cit14","first-page":"1263","article-title":"Neural message passing for quantum chemistry","volume-title":"International conference on machine learning","author":"Gilmer","year":"2017"},{"issue":"6","key":"2026101208080962500_cit15","doi-asserted-by":"publisher","first-page":"2529","DOI":"10.1021\/acs.jcim.9b00286","article-title":"RDChiral:\nAn RDKit wrapper for handling\nstereochemistry in retrosynthetic template extraction and application","volume":"59","author":"Coley","year":"2019","journal-title":"J. Chem. Inf. Model"},{"key":"2026101208080962500_cit16","doi-asserted-by":"publisher","first-page":"241","DOI":"10.1039\/D5DD00367A","article-title":"SynCat:\nmolecule-level attention graph neural network for precise reaction\nclassification","volume":"5","author":"Van Nguyen","year":"2026","journal-title":"Digital Discovery"},{"key":"2026101208080962500_cit17","doi-asserted-by":"crossref","DOI":"10.1609\/aaai.v32i1.11604","article-title":"Deeper insights into graph convolutional\nnetworks for semi-supervised learning","volume-title":"Proceedings of the AAAI conference on artificial intelligence","author":"Li","year":"2018"},{"issue":"15","key":"2026101208080962500_cit18","doi-asserted-by":"publisher","DOI":"10.1126\/sciadv.abe4166","article-title":"Extraction of organic chemistry grammar from unsupervised learning\nof chemical reactions","volume":"7","author":"Schwaller","year":"2021","journal-title":"Sci. Adv"},{"key":"2026101208080962500_cit19","unstructured":"Landrum, G.\n          \n          RDKit: Open-source\ncheminformatics. http:\/\/www.rdkit.org."},{"key":"2026101208080962500_cit20","article-title":"How\npowerful are graph neural\nnetworks?","author":"Xu","year":"2018","journal-title":"arXiv"},{"key":"2026101208080962500_cit21","first-page":"770","article-title":"Deep residual learning for\nimage recognition","volume-title":"Proceedings of the IEEE\nconference on computer vision\nand pattern recognition","author":"He","year":"2016"},{"key":"2026101208080962500_cit22","doi-asserted-by":"crossref","first-page":"3651","DOI":"10.18653\/v1\/P19-1356","article-title":"What does BERT learn about\nthe structure of language?","volume-title":"Proceedings\nof the 57th annual meeting of the association for computational linguistics","author":"Jawahar","year":"2019"},{"key":"2026101208080962500_cit23","doi-asserted-by":"crossref","first-page":"4593","DOI":"10.18653\/v1\/P19-1452","article-title":"BERT rediscovers the classical\nNLP pipeline","volume-title":"Proceedings of the 57th annual\nmeeting of the association for computational linguistics","author":"Tenney","year":"2019"},{"issue":"9","key":"2026101208080962500_cit24","doi-asserted-by":"publisher","first-page":"547","DOI":"10.3390\/a18090547","article-title":"Learning SMILES Semantics:\nWord2Vec and Transformer Embeddings for Molecular Property Prediction","volume":"18","author":"Hashemian","year":"2025","journal-title":"Algorithms"},{"issue":"1","key":"2026101208080962500_cit25","doi-asserted-by":"publisher","first-page":"964","DOI":"10.1038\/s41467-022-28536-w","article-title":"Biocatalysed synthesis\nplanning using data-driven learning","volume":"13","author":"Probst","year":"2022","journal-title":"Nat. Commun"},{"key":"2026101208080962500_cit26","volume-title":"Extraction of chemical\nstructures and reactions from the literature","author":"Lowe","year":"2012"},{"issue":"2","key":"2026101208080962500_cit27","doi-asserted-by":"publisher","first-page":"370","DOI":"10.1039\/C8SC04228D","article-title":"A graph-convolutional\nneural network model\nfor the prediction of chemical reactivity","volume":"10","author":"Coley","year":"2019","journal-title":"Chem.\nSci."},{"issue":"2","key":"2026101208080962500_cit28","doi-asserted-by":"publisher","first-page":"127","DOI":"10.1038\/s42256-021-00438-4","article-title":"Geometry-enhanced molecular representation\nlearning for property prediction","volume":"4","author":"Fang","year":"2022","journal-title":"Nat. Mach.\nIntell."},{"key":"2026101208080962500_cit29","volume-title":"Out of distribution generalization\nin machine learning","author":"Arjovsky","year":"2020"},{"issue":"1","key":"2026101208080962500_cit30","doi-asserted-by":"publisher","first-page":"471","DOI":"10.1146\/annurev-biochem-030409-143718","article-title":"Enzyme\npromiscuity: a mechanistic and evolutionary\nperspective","volume":"79","author":"Khersonsky","year":"2010","journal-title":"Annu. Rev. Biochem"},{"issue":"6","key":"2026101208080962500_cit31","doi-asserted-by":"publisher","first-page":"1185","DOI":"10.1016\/j.cell.2016.02.004","article-title":"Engineering cellular\nmetabolism","volume":"164","author":"Nielsen","year":"2016","journal-title":"Cell"},{"issue":"1","key":"2026101208080962500_cit32","doi-asserted-by":"publisher","first-page":"74","DOI":"10.1186\/s13321-024-00869-2","article-title":"AutoTemplate: enhancing\nchemical reaction datasets for machine learning\napplications in organic chemistry","volume":"16","author":"Chen","year":"2024","journal-title":"J. Cheminf"},{"issue":"2","key":"2026101208080962500_cit33","doi-asserted-by":"publisher","first-page":"1210","DOI":"10.1021\/acscatal.9b04321","article-title":"Machine learning\nin enzyme engineering","volume":"10","author":"Mazurenko","year":"2020","journal-title":"ACS Catal"},{"issue":"8016","key":"2026101208080962500_cit34","doi-asserted-by":"publisher","first-page":"493","DOI":"10.1038\/s41586-024-07487-w","article-title":"Accurate structure prediction\nof biomolecular interactions with AlphaFold 3","volume":"630","author":"Abramson","year":"2024","journal-title":"Nature"},{"issue":"6","key":"2026101208080962500_cit35","doi-asserted-by":"publisher","first-page":"483","DOI":"10.1038\/s44160-022-00231-0","article-title":"The rise of self-driving labs in\nchemical and materials\nsciences","volume":"2","author":"Abolhasani","year":"2023","journal-title":"Nat. Synth"},{"issue":"4","key":"2026101208080962500_cit36","doi-asserted-by":"publisher","first-page":"1085","DOI":"10.1093\/bib\/bbx085","article-title":"The BioCyc collection\nof microbial genomes and metabolic pathways","volume":"20","author":"Karp","year":"2019","journal-title":"Briefings Bioinf."}],"container-title":["Journal of Chemical Information\nand Modeling"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/pubs.acs.org\/doi\/pdf\/10.1021\/acs.jcim.6c02672","content-type":"application\/pdf","content-version":"vor","intended-application":"unspecified"},{"URL":"https:\/\/pubs.acs.org\/jcisd8\/article-pdf\/66\/19\/12904\/68714913\/acs.jcim.6c02672.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/pubs.acs.org\/jcisd8\/article-pdf\/66\/19\/12904\/68714913\/acs.jcim.6c02672.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,10,12]],"date-time":"2026-10-12T12:50:01Z","timestamp":1791809401000},"score":1,"resource":{"primary":{"URL":"https:\/\/pubs.acs.org\/jcisd8\/article\/66\/19\/12904\/5442605\/GMSF-A-Dual-Path-Multimodal-Framework-for-Enzyme"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,9,30]]},"references-count":36,"journal-issue":{"issue":"19","published-online":{"date-parts":[[2026,9,30]]},"published-print":{"date-parts":[[2026,10,12]]}},"URL":"https:\/\/doi.org\/10.1021\/acs.jcim.6c02672","relation":{},"ISSN":["1549-9596","1549-960X"],"issn-type":[{"value":"1549-9596","type":"print"},{"value":"1549-960X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,9,30]]}}}