{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,15]],"date-time":"2026-10-15T05:48:41Z","timestamp":1792043321527,"version":"4.4.0"},"reference-count":26,"publisher":"Oxford University Press (OUP)","issue":"10","license":[{"start":{"date-parts":[[2026,10,2]],"date-time":"2026-10-02T00:00:00Z","timestamp":1790899200000},"content-version":"vor","delay-in-days":1,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2026,10,2]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:sec>\n                    <jats:title>Motivation<\/jats:title>\n                    <jats:p>Enzymes are essential catalysts in many cellular processes. Understanding their interactions with small molecules, such as regulators, cofactors, and most importantly, substrates, is crucial for understanding the biochemical processes that occur in cells. Correctly interpreting the roles of small molecules that interact with enzymes is key to elucidating enzyme function. Recently, enzyme\u2013small-molecule interaction prediction has attracted growing interest from computational methods, especially deep learning. As a result, researchers have published several datasets and numerous models with remarkable performance.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>In this work, we critically examine one of the most popular datasets and four models trained on it, identifying similarity-induced information leakage that may overinflate reported model performance in out-of-distribution applications. We show that the inspected models are susceptible to information leakage, and their performance is not better than statistical baselines when the leakage is removed. Furthermore, controlling for leakage due to protein sequence similarity alone still yields high predictive performance, whereas performance decreases substantially when common substrates shared between training and test sets are removed from one of them, and approaches random prediction when structurally similar ligands are separated between splits. Thus, the investigated models generalize considerably better across enzyme sequence space than across ligand chemical space, indicating that their reported performance depends strongly on the small-molecule similarity between training and evaluation data.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Availability and implementation<\/jats:title>\n                    <jats:p>All data splits we calculated in this study are available on Zenodo (DOI: 10.5281\/zenodo.18786394). The code is available on GitHub and backed up on Zenodo (DOI: 10.5281\/zenodo.18788609).<\/jats:p>\n                  <\/jats:sec>","DOI":"10.1093\/bioinformatics\/btag733","type":"journal-article","created":{"date-parts":[[2026,9,30]],"date-time":"2026-09-30T12:02:35Z","timestamp":1790769755000},"source":"Crossref","is-referenced-by-count":0,"title":["Information leakage in enzyme-substrate prediction"],"prefix":"10.1093","volume":"42","author":[{"given":"Vahid","family":"Atabaigi Elmi","sequence":"first","affiliation":[{"name":"Drug Bioinformatics, Helmholtz Institute for Pharmaceutical Research Saarland , Saarbruecken, Saarland 66123,","place":["Germany"]},{"name":"Centre for Bioinformatics, Saarland University , Saarbruecken, Saarland 66123,","place":["Germany"]},{"name":"Department of Chemistry and Molecular Biology, University of Gothenburg , Gothenburg 41390,","place":["Sweden"]},{"name":"Wallenberg Centre for Molecular and Translational Medicine, University of Gothenburg , Gothenburg 41390,","place":["Sweden"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Roman","family":"Joeres","sequence":"additional","affiliation":[{"name":"Drug Bioinformatics, Helmholtz Institute for Pharmaceutical Research Saarland , Saarbruecken, Saarland 66123,","place":["Germany"]},{"name":"Centre for Bioinformatics, Saarland University , Saarbruecken, Saarland 66123,","place":["Germany"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Olga V","family":"Kalinina","sequence":"additional","affiliation":[{"name":"Drug Bioinformatics, Helmholtz Institute for Pharmaceutical Research Saarland , Saarbruecken, Saarland 66123,","place":["Germany"]},{"name":"Centre for Bioinformatics, Saarland University , Saarbruecken, Saarland 66123,","place":["Germany"]},{"name":"Medical Faculty, Saarland University , Homburg, Saarland 66421,","place":["Germany"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"286","published-online":{"date-parts":[[2026,10,1]]},"reference":[{"key":"2026101501272282000_btag733-B2","doi-asserted-by":"publisher","first-page":"1555","DOI":"10.1093\/jxb\/ern091","article-title":"Catalysis and regulation in rubisco","volume":"59","author":"Andersson","year":"2008","journal-title":"J Exp Bot"},{"key":"2026101501272282000_btag733-B3","volume-title":"Biochemistry","author":"Berg","year":"2012"},{"key":"2026101501272282000_btag733-B4","doi-asserted-by":"publisher","first-page":"bbae076","DOI":"10.1093\/bib\/bbae076","article-title":"Cracking the black box of deep sequence-based protein\u2013protein interaction prediction","volume":"25","author":"Bernett","year":"2024","journal-title":"Brief Bioinform"},{"key":"2026101501272282000_btag733-B5","doi-asserted-by":"publisher","first-page":"1444","DOI":"10.1038\/s41592-024-02362-y","article-title":"Guiding questions to avoid data leakage in biological machine learning applications","volume":"21","author":"Bernett","year":"2024","journal-title":"Nat Methods"},{"key":"2026101501272282000_btag733-B6","doi-asserted-by":"publisher","first-page":"59","DOI":"10.1038\/nmeth.3176","article-title":"Fast and sensitive protein alignment using DIAMOND","volume":"12","author":"Buchfink","year":"2015","journal-title":"Nat Methods"},{"key":"2026101501272282000_btag733-B7","doi-asserted-by":"publisher","first-page":"407","DOI":"10.1021\/ar5002928","article-title":"The dynamical nature of enzymatic catalysis","volume":"48","author":"Callender","year":"2015","journal-title":"Acc Chem Res"},{"key":"2026101501272282000_btag733-B8","doi-asserted-by":"publisher","first-page":"1989","DOI":"10.1038\/s41467-023-37572-z","article-title":"Improving the generalizability of protein-ligand binding predictions with AI-Bind","volume":"14","author":"Chatterjee","year":"2023","journal-title":"Nat Commun"},{"key":"2026101501272282000_btag733-B9","doi-asserted-by":"publisher","author":"Corso","year":"2022","DOI":"10.48550\/arXiv.2210.01776"},{"key":"2026101501272282000_btag733-B10","doi-asserted-by":"publisher","first-page":"639","DOI":"10.1038\/s41586-025-09697-2","article-title":"Enzyme specificity prediction using cross attention graph neural networks","volume":"647","author":"Cui","year":"2025","journal-title":"Nature"},{"key":"2026101501272282000_btag733-B12","doi-asserted-by":"publisher","first-page":"2806","DOI":"10.1021\/acs.jcim.4c02357","article-title":"FusionESP: improved enzyme\u2013substrate pair prediction by fusing protein and chemical knowledge","volume":"65","author":"Du","year":"2025","journal-title":"J Chem Inf Model"},{"key":"2026101501272282000_btag733-B13","doi-asserted-by":"publisher","first-page":"10422","DOI":"10.1021\/bi201486f","article-title":"Flexibility, diversity, and cooperativity: pillars of enzyme catalysis","volume":"50","author":"Hammes","year":"2011","journal-title":"Biochemistry"},{"key":"2026101501272282000_btag733-B14","doi-asserted-by":"publisher","first-page":"3337","DOI":"10.1038\/s41467-025-58606-8","article-title":"Data splitting to avoid information leakage with DataSAIL","volume":"16","author":"Joeres","year":"2025","journal-title":"Nat Commun"},{"key":"2026101501272282000_btag733-B15","doi-asserted-by":"publisher","first-page":"471","DOI":"10.1146\/annurev-biochem-030409-143718","article-title":"Enzyme promiscuity: a mechanistic and evolutionary perspective","volume":"79","author":"Khersonsky","year":"2010","journal-title":"Annu Rev Biochem"},{"key":"2026101501272282000_btag733-B17","doi-asserted-by":"publisher","first-page":"2787","DOI":"10.1038\/s41467-023-38347-2","article-title":"A general model to predict small molecule substrates of enzymes based on machine and deep learning","volume":"14","author":"Kroll","year":"2023","journal-title":"Nat Commun"},{"key":"2026101501272282000_btag733-B18","doi-asserted-by":"publisher","first-page":"e1012100","DOI":"10.1371\/journal.pcbi.1012100","article-title":"A multimodal transformer network for protein-small molecule interactions enhances predictions of kinase inhibition and enzyme-substrate relationships","volume":"20","author":"Kroll","year":"2024","journal-title":"PLoS Comput Biol"},{"key":"2026101501272282000_btag733-B19","doi-asserted-by":"publisher","first-page":"70","DOI":"10.1016\/j.bbapap.2015.08.011","author":"Kuo","year":"2016"},{"key":"2026101501272282000_btag733-B20","doi-asserted-by":"publisher","first-page":"135","DOI":"10.1186\/s13321-026-01287-2","article-title":"DockM8: an all-in-one open-source platform for consensus virtual screening in drug design","volume":"18","author":"Lacour","year":"2026","journal-title":"J Cheminform"},{"key":"2026101501272282000_btag733-B21","doi-asserted-by":"publisher","first-page":"1658","DOI":"10.1093\/bioinformatics\/btl158","article-title":"Cd-hit: a fast program for clustering and comparing large sets of protein or nucleotide sequences","volume":"22","author":"Li","year":"2006","journal-title":"Bioinformatics"},{"key":"2026101501272282000_btag733-B24","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2506.17963","article-title":"OmniESI: a unified framework for enzyme-substrate interaction prediction with progressive conditional deep learning","author":"Nie","year":"2025"},{"key":"2026101501272282000_btag733-B25","doi-asserted-by":"publisher","first-page":"171","DOI":"10.1016\/j.ymben.2017.09.016","article-title":"Predicting novel substrates for enzymes with minimal experimental effort with active learning","volume":"44","author":"Pertusi","year":"2017","journal-title":"Metab Eng"},{"key":"2026101501272282000_btag733-B26","doi-asserted-by":"publisher","first-page":"41","DOI":"10.1038\/nrd.2018.168","article-title":"Drug repurposing: progress, challenges and recommendations","volume":"18","author":"Pushpakom","year":"2019","journal-title":"Nat Rev Drug Discov"},{"key":"2026101501272282000_btag733-B27","doi-asserted-by":"publisher","first-page":"187","DOI":"10.1021\/acs.jcim.4c01801","article-title":"Deep Learning-Driven insights into enzyme\u2013substrate interaction discovery","volume":"65","author":"Qian","year":"2025","journal-title":"J Chem Inf Model"},{"key":"2026101501272282000_btag733-B28","doi-asserted-by":"publisher","first-page":"i590","DOI":"10.1093\/bioinformatics\/btaf192","article-title":"Deep learning models for unbiased sequence-based PPI prediction Plateau at an accuracy of 0.65","volume":"41","author":"Reim","year":"2025","journal-title":"Bioinformatics"},{"key":"2026101501272282000_btag733-B30","doi-asserted-by":"publisher","first-page":"167517","DOI":"10.1016\/j.jmb.2022.167517","article-title":"Conformational variation in enzyme catalysis: a structural study on catalytic residues","volume":"434","author":"Riziotis","year":"2022","journal-title":"J Mol Biol"},{"key":"2026101501272282000_btag733-B31","doi-asserted-by":"publisher","first-page":"455","DOI":"10.1002\/jcc.21334","article-title":"AutoDock vina: improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading","volume":"31","author":"Trott","year":"2010","journal-title":"J Comput Chem"},{"key":"2026101501272282000_btag733-B33","doi-asserted-by":"publisher","first-page":"1459954","DOI":"10.3389\/fphar.2024.1459954","article-title":"The role of artificial intelligence in drug screening, drug design, and clinical trials","volume":"15","author":"Wu","year":"2024","journal-title":"Front Pharmacol"}],"container-title":["Bioinformatics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/academic.oup.com\/bioinformatics\/advance-article-pdf\/doi\/10.1093\/bioinformatics\/btag733\/71575338\/btag733.pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/bioinformatics\/article-pdf\/42\/10\/btag733\/71575338\/btag733.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/bioinformatics\/article-pdf\/42\/10\/btag733\/71575338\/btag733.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,10,15]],"date-time":"2026-10-15T05:27:31Z","timestamp":1792042051000},"score":1,"resource":{"primary":{"URL":"https:\/\/academic.oup.com\/bioinformatics\/article\/doi\/10.1093\/bioinformatics\/btag733\/8860137"}},"subtitle":[],"editor":[{"given":"Lenore","family":"Cowen","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"editor"}]}],"short-title":[],"issued":{"date-parts":[[2026,10,1]]},"references-count":26,"journal-issue":{"issue":"10","published-print":{"date-parts":[[2026,10,2]]}},"URL":"https:\/\/doi.org\/10.1093\/bioinformatics\/btag733","relation":{},"ISSN":["1367-4803","1367-4811"],"issn-type":[{"value":"1367-4803","type":"print"},{"value":"1367-4811","type":"electronic"}],"subject":[],"published-other":{"date-parts":[[2026,10]]},"published":{"date-parts":[[2026,10,1]]},"article-number":"btag733"}}