{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,26]],"date-time":"2026-06-26T09:32:24Z","timestamp":1782466344007,"version":"3.54.5"},"reference-count":16,"publisher":"Oxford University Press (OUP)","issue":"2","license":[{"start":{"date-parts":[[2024,1,20]],"date-time":"2024-01-20T00:00:00Z","timestamp":1705708800000},"content-version":"vor","delay-in-days":1,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/100000133","name":"Agency for Healthcare Research and Quality","doi-asserted-by":"publisher","award":["R01HS029009"],"award-info":[{"award-number":["R01HS029009"]}],"id":[{"id":"10.13039\/100000133","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100000002","name":"National Institutes of Health","doi-asserted-by":"publisher","award":["R35GM139656"],"award-info":[{"award-number":["R35GM139656"]}],"id":[{"id":"10.13039\/100000002","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2024,2,1]]},"abstract":"<jats:title>Abstract<\/jats:title>\n               <jats:sec>\n                  <jats:title>Motivation<\/jats:title>\n                  <jats:p>Phosphorylation, a post-translational modification regulated by protein kinase enzymes, plays an essential role in almost all cellular processes. Understanding how each of the nearly 500 human protein kinases selectively phosphorylates their substrates is a foundational challenge in bioinformatics and cell signaling. Although deep learning models have been a popular means to predict kinase\u2013substrate relationships, existing models often lack interpretability and are trained on datasets skewed toward a subset of well-studied kinases.<\/jats:p>\n               <\/jats:sec>\n               <jats:sec>\n                  <jats:title>Results<\/jats:title>\n                  <jats:p>Here we leverage recent peptide library datasets generated to determine substrate specificity profiles of 300 serine\/threonine kinases to develop an explainable Transformer model for kinase\u2013peptide interaction prediction. The model, trained solely on primary sequences, achieved state-of-the-art performance. Its unique multitask learning paradigm built within the model enables predictions on virtually any kinase\u2013peptide pair, including predictions on 139 kinases not used in peptide library screens. Furthermore, we employed explainable machine learning methods to elucidate the model\u2019s inner workings. Through analysis of learned embeddings at different training stages, we demonstrate that the model employs a unique strategy of substrate prediction considering both substrate motif patterns and kinase evolutionary features. SHapley Additive exPlanation (SHAP) analysis reveals key specificity determining residues in the peptide sequence. Finally, we provide a web interface for predicting kinase\u2013substrate associations for user-defined sequences and a resource for visualizing the learned kinase\u2013substrate associations.<\/jats:p>\n               <\/jats:sec>\n               <jats:sec>\n                  <jats:title>Availability and implementation<\/jats:title>\n                  <jats:p>All code and data are available at https:\/\/github.com\/esbgkannan\/Phosformer-ST. Web server is available at https:\/\/phosformer.netlify.app.<\/jats:p>\n               <\/jats:sec>","DOI":"10.1093\/bioinformatics\/btae033","type":"journal-article","created":{"date-parts":[[2024,1,21]],"date-time":"2024-01-21T01:14:27Z","timestamp":1705799667000},"source":"Crossref","is-referenced-by-count":17,"title":["Using explainable machine learning to uncover the kinase\u2013substrate interaction landscape"],"prefix":"10.1093","volume":"40","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-4471-6759","authenticated-orcid":false,"given":"Zhongliang","family":"Zhou","sequence":"first","affiliation":[{"name":"School of Computing, University of Georgia , Athens, GA 30602, United States"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Wayland","family":"Yeung","sequence":"additional","affiliation":[{"name":"Institute of Bioinformatics, University of Georgia , Athens, GA 30602, United States"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Saber","family":"Soleymani","sequence":"additional","affiliation":[{"name":"School of Computing, University of Georgia , Athens, GA 30602, United States"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Nathan","family":"Gravel","sequence":"additional","affiliation":[{"name":"Institute of Bioinformatics, University of Georgia , Athens, GA 30602, United States"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Mariah","family":"Salcedo","sequence":"additional","affiliation":[{"name":"Department of Biochemistry and Molecular Biology, University of Georgia , Athens, GA 30602, United States"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Sheng","family":"Li","sequence":"additional","affiliation":[{"name":"School of Data Science, University of Virginia , Charlottesville, VA 22903, United States"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-2833-8375","authenticated-orcid":false,"given":"Natarajan","family":"Kannan","sequence":"additional","affiliation":[{"name":"Institute of Bioinformatics, University of Georgia , Athens, GA 30602, United States"},{"name":"Department of Biochemistry and Molecular Biology, University of Georgia , Athens, GA 30602, United States"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"286","published-online":{"date-parts":[[2024,1,19]]},"reference":[{"key":"2024020905541843900_btae033-B1","doi-asserted-by":"crossref","first-page":"610","DOI":"10.1016\/j.devcel.2019.06.008","article-title":"Phosphorylation of Ci\/Gli by fused family kinases promotes hedgehog signaling","volume":"50","author":"Han","year":"2019","journal-title":"Dev Cell"},{"key":"2024020905541843900_btae033-B2","doi-asserted-by":"crossref","first-page":"576","DOI":"10.1096\/fasebj.9.8.7768349","article-title":"The eukaryotic protein kinase superfamily: kinase (catalytic) domain structure and classification 1","volume":"9","author":"Hanks","year":"1995","journal-title":"FASEB J"},{"key":"2024020905541843900_btae033-B3","doi-asserted-by":"crossref","first-page":"D261","DOI":"10.1093\/nar\/gkr1122","article-title":"Phosphositeplus: a comprehensive resource for investigating the structure and function of experimentally determined post-translational modifications in man and mouse","volume":"40","author":"Hornbeck","year":"2012","journal-title":"Nucleic Acids Res"},{"key":"2024020905541843900_btae033-B4","doi-asserted-by":"crossref","first-page":"759","DOI":"10.1038\/s41586-022-05575-3","article-title":"An atlas of substrate specificities for the human serine\/threonine kinome","volume":"613","author":"Johnson","year":"2023","journal-title":"Nature"},{"key":"2024020905541843900_btae033-B5","doi-asserted-by":"crossref","first-page":"2119","DOI":"10.1093\/bioinformatics\/btac083","article-title":"Ember: multi-label prediction of kinase-substrate phosphorylation events through deep learning","volume":"38","author":"Kirchoff","year":"2022","journal-title":"Bioinformatics"},{"key":"2024020905541843900_btae033-B6","first-page":"4768","article-title":"A unified approach to interpreting model predictions","volume":"30","author":"Lundberg","year":"2017","journal-title":"Adv Neural Inf Process Syst"},{"key":"2024020905541843900_btae033-B7","doi-asserted-by":"crossref","first-page":"2766","DOI":"10.1093\/bioinformatics\/bty1051","article-title":"DeepPhos: prediction of protein phosphorylation sites with deep learning","volume":"35","author":"Luo","year":"2019","journal-title":"Bioinformatics"},{"key":"2024020905541843900_btae033-B8","doi-asserted-by":"crossref","first-page":"1912","DOI":"10.1126\/science.1075762","article-title":"The protein kinase complement of the human genome","volume":"298","author":"Manning","year":"2002","journal-title":"Science"},{"key":"2024020905541843900_btae033-B9","first-page":"861","volume-title":"J Open Source Softw","author":"McInnes","year":"2018"},{"key":"2024020905541843900_btae033-B10","doi-asserted-by":"crossref","first-page":"191","DOI":"10.1016\/S0167-4889(96)00083-3","article-title":"How do protein kinases recognize their substrates?","volume":"1314","author":"Pinna","year":"1996","journal-title":"Biochim Biophys Acta"},{"key":"2024020905541843900_btae033-B11","doi-asserted-by":"crossref","first-page":"e2016239118","DOI":"10.1073\/pnas.2016239118","article-title":"Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences","volume":"118","author":"Rives","year":"2021","journal-title":"Proc Natl Acad Sci USA"},{"key":"2024020905541843900_btae033-B12","doi-asserted-by":"crossref","first-page":"647","DOI":"10.1007\/s10115-013-0679-x","article-title":"Explaining prediction models and individual predictions with feature contributions","volume":"41","author":"\u0160trumbelj","year":"2014","journal-title":"Knowl Inf Syst"},{"key":"2024020905541843900_btae033-B13","doi-asserted-by":"crossref","first-page":"3909","DOI":"10.1093\/bioinformatics\/btx496","article-title":"MusiteDeep: a deep-learning framework for general and kinase-specific phosphorylation site prediction","volume":"33","author":"Wang","year":"2017","journal-title":"Bioinformatics"},{"key":"2024020905541843900_btae033-B14","doi-asserted-by":"crossref","first-page":"4668","DOI":"10.1093\/bioinformatics\/btab551","article-title":"PhosIDN: an integrated deep neural network for improving protein phosphorylation site prediction by combining sequence and protein\u2013protein interaction information","volume":"37","author":"Yang","year":"2021","journal-title":"Bioinformatics"},{"key":"2024020905541843900_btae033-B15","doi-asserted-by":"crossref","first-page":"btad046","DOI":"10.1093\/bioinformatics\/btad046","article-title":"Phosformer: an explainable transformer model for protein kinase-specific phosphorylation predictions","volume":"39","author":"Zhou","year":"2023","journal-title":"Bioinformatics"},{"key":"2024020905541843900_btae033-B16","doi-asserted-by":"crossref","first-page":"36372","DOI":"10.1074\/jbc.M505031200","article-title":"A single pair of acidic residues in the kinase major groove mediates strong substrate preference for P-2 or P-5 arginine in the AGC, CAMK, and STE kinase families","volume":"280","author":"Zhu","year":"2005","journal-title":"J Biol Chem"}],"container-title":["Bioinformatics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/academic.oup.com\/bioinformatics\/advance-article-pdf\/doi\/10.1093\/bioinformatics\/btae033\/56284976\/btae033.pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/bioinformatics\/article-pdf\/40\/2\/btae033\/56630415\/btae033.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/bioinformatics\/article-pdf\/40\/2\/btae033\/56630415\/btae033.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,2,9]],"date-time":"2024-02-09T08:29:29Z","timestamp":1707467369000},"score":1,"resource":{"primary":{"URL":"https:\/\/academic.oup.com\/bioinformatics\/article\/doi\/10.1093\/bioinformatics\/btae033\/7582285"}},"subtitle":[],"editor":[{"given":"Pier Luigi","family":"Martelli","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"editor"}]}],"short-title":[],"issued":{"date-parts":[[2024,1,19]]},"references-count":16,"journal-issue":{"issue":"2","published-print":{"date-parts":[[2024,2,1]]}},"URL":"https:\/\/doi.org\/10.1093\/bioinformatics\/btae033","relation":{},"ISSN":["1367-4811"],"issn-type":[{"value":"1367-4811","type":"electronic"}],"subject":[],"published-other":{"date-parts":[[2024,2,1]]},"published":{"date-parts":[[2024,1,19]]},"article-number":"btae033"}}