{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,3]],"date-time":"2026-09-03T04:03:46Z","timestamp":1788408226081,"version":"build-2803163510"},"reference-count":23,"publisher":"Oxford University Press (OUP)","issue":"4","license":[{"start":{"date-parts":[[2026,7,7]],"date-time":"2026-07-07T00:00:00Z","timestamp":1783382400000},"content-version":"vor","delay-in-days":6,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2026,7,3]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:p>The spontaneous deamidation of asparagine residues remains a major obstacle to the stability and efficacy of protein therapeutics. Currently available models in the literature for predicting deamidation liabilities can suffer from limited generalizability, likely due to biases such as sequence similarity within datasets. In this study, we built machine learning models using protein language models (PLMs, e.g. ESM2) and graph neural networks (GNNs), trained on a comprehensive dataset of 591 asparagine sites from over 105 protein molecules. To address the critical issue of data leakage, we implemented a peptide grouping strategy yielding more accurate estimates of model performance for novel deamidation sites. Our analysis shows that, when sequence similarity bias is controlled, PLMs match traditional feature-based models that use amino acid composition and predicted secondary structure\/solvent accessibility, while offering substantial computational advantages. Additionally, our GNN-based pipeline further increases prediction accuracy by $\\sim$8% compared with language model-only tools and delivers a 15%\u201325% improvement over motif-based approaches. This methodological framework enables more reliable and rapid in-silico prediction of deamidation liabilities, potentially reducing costly late-stage interventions in protein therapeutic development and is generalizable to the modeling of other protein post-translational modifications.<\/jats:p>","DOI":"10.1093\/bib\/bbag363","type":"journal-article","created":{"date-parts":[[2026,7,7]],"date-time":"2026-07-07T12:13:30Z","timestamp":1783426410000},"source":"Crossref","is-referenced-by-count":1,"title":["Accurate prediction of asparagine deamidation in biologics using advanced machine learning models"],"prefix":"10.1093","volume":"27","author":[{"given":"Shafayat","family":"Ahmed","sequence":"first","affiliation":[{"name":"Data Science and Modelling, Biopharmaceutical Development, AstraZeneca , 1 MedImmune Way, Gaithersburg, MD 20878 ,","place":["United States"]},{"name":"Department of Computer Science, Virginia Tech , 620 Drillfield Dr., Blacksburg, VA 24060 ,","place":["United States"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Nicole","family":"Swope","sequence":"additional","affiliation":[{"name":"Analytical Sciences, Biopharmaceutical Development, AstraZeneca , 1 MedImmune Way, Gaithersburg, MD 20878 ,","place":["United States"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Valentin","family":"Stanev","sequence":"additional","affiliation":[{"name":"Data Science and Modelling, Biopharmaceutical Development, AstraZeneca , 1 MedImmune Way, Gaithersburg, MD 20878 ,","place":["United States"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Romina","family":"Hofele","sequence":"additional","affiliation":[{"name":"Analytical Sciences, Biopharmaceutical Development, AstraZeneca , 1 MedImmune Way, Gaithersburg, MD 20878 ,","place":["United States"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Dominique","family":"WuDunn","sequence":"additional","affiliation":[{"name":"Analytical Sciences, Biopharmaceutical Development, AstraZeneca , 1 MedImmune Way, Gaithersburg, MD 20878 ,","place":["United States"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Rohan","family":"Jain","sequence":"additional","affiliation":[{"name":"Data Science and Modelling, Biopharmaceutical Development, AstraZeneca , 1 MedImmune Way, Gaithersburg, MD 20878 ,","place":["United States"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0004-2642-7253","authenticated-orcid":false,"given":"Jared","family":"Delmar","sequence":"additional","affiliation":[{"name":"Analytical Sciences, Biopharmaceutical Development, AstraZeneca , 1 MedImmune Way, Gaithersburg, MD 20878 ,","place":["United States"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-0445-8364","authenticated-orcid":false,"given":"Maryam","family":"Pouryahya","sequence":"additional","affiliation":[{"name":"Data Science and Modelling, Biopharmaceutical Development, AstraZeneca , 1 MedImmune Way, Gaithersburg, MD 20878 ,","place":["United States"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"286","published-online":{"date-parts":[[2026,7,7]]},"reference":[{"key":"2026070708092929900_ref1","volume-title":"Drugs@FDA: FDA Approved Drug Products","author":"U.S. Food and Drug Administration","year":"2025"},{"key":"2026070708092929900_ref2","doi-asserted-by":"publisher","first-page":"944","DOI":"10.1073\/pnas.98.3.944","article-title":"Molecular clocks","volume":"98","author":"Robinson","year":"2001","journal-title":"Proc Natl Acad Sci U S A"},{"key":"2026070708092929900_ref3","doi-asserted-by":"publisher","first-page":"1111","DOI":"10.1021\/jp808597v","article-title":"Deamidation of asparagine residues: direct hydrolysis versus succinimide-mediated deamidation mechanisms","volume":"113","author":"Catak","year":"2009","journal-title":"J Phys Chem A"},{"key":"2026070708092929900_ref4","doi-asserted-by":"publisher","first-page":"327","DOI":"10.4161\/mabs.27876","article-title":"Assessment of chemical modifications of sites in the CDRs of recombinant antibodies: susceptibility vs. functionality of critical quality attributes","volume":"6","author":"Haberger","year":"2014","journal-title":"MAbs"},{"key":"2026070708092929900_ref5","doi-asserted-by":"publisher","first-page":"426","DOI":"10.1111\/j.1399-3011.2004.00151.x","article-title":"Structure-dependent nonenzymatic deamidation of glutaminyl and asparaginyl pentapeptides","volume":"63","author":"Robinson","year":"2004","journal-title":"J Pept Res"},{"key":"2026070708092929900_ref6","doi-asserted-by":"publisher","first-page":"e0181347","DOI":"10.1371\/journal.pone.0181347","article-title":"Protein asparagine deamidation prediction based on structures with machine learning methods","volume":"12","author":"Jia","year":"2017","journal-title":"PLoS One"},{"key":"2026070708092929900_ref7","doi-asserted-by":"publisher","first-page":"901","DOI":"10.1080\/19420862.2018.1478646","article-title":"Structure based prediction of asparagine deamidation propensity in monoclonal antibodies","volume":"10","author":"Yan","year":"2018","journal-title":"MAbs"},{"key":"2026070708092929900_ref8","doi-asserted-by":"publisher","first-page":"4367","DOI":"10.1073\/pnas.071066498","article-title":"Prediction of protein deamidation rates from primary and three-dimensional structure","volume":"98","author":"Robinson","year":"2001","journal-title":"Proc Natl Acad Sci U S A"},{"key":"2026070708092929900_ref9","doi-asserted-by":"publisher","first-page":"e0145186","DOI":"10.1371\/journal.pone.0145186","article-title":"Prediction of spontaneous protein deamidation from sequence-derived secondary structure and intrinsic disorder","volume":"10","author":"Lorenzo","year":"2015","journal-title":"PLoS One"},{"key":"2026070708092929900_ref10","doi-asserted-by":"publisher","first-page":"e100736","DOI":"10.1371\/journal.pone.0100736","article-title":"Structure-based prediction of asparagine and aspartate degradation sites in antibody variable regions","volume":"9","author":"Sydow","year":"2014","journal-title":"PLoS One"},{"key":"2026070708092929900_ref11","doi-asserted-by":"publisher","first-page":"264","DOI":"10.1016\/j.omtm.2019.09.008","article-title":"Machine learning enables accurate prediction of asparagine deamidation probability and rate","volume":"15","author":"Delmar","year":"2019","journal-title":"Mol Ther Methods Clin Dev"},{"key":"2026070708092929900_ref12","doi-asserted-by":"publisher","first-page":"2333436","DOI":"10.1080\/19420862.2024.2333436","article-title":"Predicting deamidation and isomerization sites in therapeutic antibodies using structure-based in silico approaches","volume":"16","author":"Hoffmann","year":"2024","journal-title":"MAbs"},{"key":"2026070708092929900_ref13","doi-asserted-by":"publisher","first-page":"e1011939","DOI":"10.1371\/journal.pcbi.1011939","article-title":"Combining machine learning with structure-based protein design to predict and engineer post-translational modifications of proteins","volume":"20","author":"Ertelt","year":"2024","journal-title":"PLoS Comput Biol"},{"key":"2026070708092929900_ref14","doi-asserted-by":"publisher","first-page":"583","DOI":"10.1038\/s41586-021-03819-2","article-title":"Highly accurate protein structure prediction with AlphaFold","volume":"596","author":"Jumper","year":"2021","journal-title":"Nature"},{"key":"2026070708092929900_ref15","doi-asserted-by":"publisher","first-page":"74","DOI":"10.3390\/antib13030074","article-title":"The accurate prediction of antibody deamidations by combining high-throughput automated peptide mapping and protein language model-based deep learning","volume":"13","author":"Niu","year":"2024","journal-title":"Antibodies (Basel)"},{"key":"2026070708092929900_ref16","doi-asserted-by":"publisher","first-page":"466","DOI":"10.1016\/j.omtm.2021.03.023","article-title":"Machine learning prediction of methionine and tryptophan photooxidation susceptibility","volume":"21","author":"Delmar","year":"2021","journal-title":"Mol Ther Methods Clin Dev"},{"key":"2026070708092929900_ref17","doi-asserted-by":"publisher","DOI":"10.1101\/2021.10.04.463034","article-title":"Protein complex prediction with AlphaFold-Multimer","volume-title":"bioRxiv","author":"Evans","year":"2021"},{"key":"2026070708092929900_ref18","doi-asserted-by":"publisher","first-page":"426","DOI":"10.1111\/j.1399-3011.2004.00151.x","article-title":"Structure-dependent nonenzymatic deamidation of glutaminyl and asparaginyl pentapeptides","volume":"63","author":"Robinson","year":"2004","journal-title":"J Pept Res"},{"key":"2026070708092929900_ref19","doi-asserted-by":"publisher","first-page":"17","DOI":"10.1002\/jmr.647","article-title":"IMGT standardized criteria for statistical analysis of immunoglobulin V-REGION amino acid properties","volume":"17","author":"Pommi\u00e9","year":"2004","journal-title":"J Mol Recognit"},{"key":"2026070708092929900_ref20","doi-asserted-by":"publisher","first-page":"6004","DOI":"10.1021\/ac050672d","article-title":"Identification and characterization of Deamidation sites in the conserved regions of human immunoglobulin gamma antibodies","volume":"77","author":"Chelius","year":"2005","journal-title":"Anal Chem"},{"key":"2026070708092929900_ref21","first-page":"2825","article-title":"Scikit-learn: machine learning in python","volume":"12","author":"Pedregosa","year":"2011","journal-title":"J Mach Learn Res"},{"key":"2026070708092929900_ref22","doi-asserted-by":"publisher","DOI":"10.1101\/2022.07.20.500902","article-title":"Language models of protein sequences at the scale of evolution enable accurate structure prediction","volume-title":"bioRxiv","author":"Lin"},{"key":"2026070708092929900_ref23","doi-asserted-by":"publisher","first-page":"2562997","DOI":"10.1080\/19420862.2025.2562997","article-title":"Accelerating antibody development: sequence and structure-based models for predicting developability properties via size exclusion chromatography","volume":"17","author":"Abeer","year":"2025","journal-title":"MAbs"}],"container-title":["Briefings in Bioinformatics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/academic.oup.com\/bib\/article-pdf\/27\/4\/bbag363\/68707056\/bbag363.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/bib\/article-pdf\/27\/4\/bbag363\/68707056\/bbag363.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,7,7]],"date-time":"2026-07-07T12:13:32Z","timestamp":1783426412000},"score":1,"resource":{"primary":{"URL":"https:\/\/academic.oup.com\/bib\/article\/doi\/10.1093\/bib\/bbag363\/8726037"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,7]]},"references-count":23,"journal-issue":{"issue":"4","published-print":{"date-parts":[[2026,7,3]]}},"URL":"https:\/\/doi.org\/10.1093\/bib\/bbag363","relation":{},"ISSN":["1467-5463","1477-4054"],"issn-type":[{"value":"1467-5463","type":"print"},{"value":"1477-4054","type":"electronic"}],"subject":[],"published-other":{"date-parts":[[2026,7]]},"published":{"date-parts":[[2026,7]]},"article-number":"bbag363"}}