{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,10]],"date-time":"2026-07-10T13:14:29Z","timestamp":1783689269382,"version":"3.55.0"},"reference-count":49,"publisher":"Oxford University Press (OUP)","issue":"Supplement_1","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\/"}],"funder":[{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["62332020"],"award-info":[{"award-number":["62332020"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"name":"Xiangjiang Laboratory","award":["23XJ01011"],"award-info":[{"award-number":["23XJ01011"]}]},{"name":"High Performance Computing Center of Central South University"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2026,7,1]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:sec>\n                    <jats:title>Motivation<\/jats:title>\n                    <jats:p>Accurately estimating changes in binding free energy (\u0394\u0394G) is critical for understanding protein\u2013protein interactions (PPIs) and guiding rational protein design. Recent deep learning methods have achieved notable progress by pre-training on large-scale structural data. While some approaches explore structural flexibility through energy-based sampling or generative modeling, these strategies typically involve substantial computational cost and overlook data efficacy in terms of training data organization.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>We present USP-ddG, a unified structural paradigm for \u0394\u0394G prediction built on a dual-channel architecture. The model incorporates three complementary components: (i) an inverse folding-based log-odds ratio, (ii) the empirical force field FoldX capturing side-chain packing energetics, and (iii) a geometric encoder that leverages Gaussian coordinate perturbation as a regularization strategy to improve robustness. To enhance representation capacity, we introduce a framework that integrates feed-forward network (FFN) and Mixture-of-Experts (MoE) to model domain-invariant and -specific features, respectively. We further propose CATH-guided Folding Ordering (CFO), a data efficacy strategy that organizes samples to mitigate catastrophic forgetting and data distribution bias. USP-ddG consistently outperforms existing state-of-the-art methods on the SKEMPI v2.0 benchmark, including the challenging hold-out CATH test set. It achieves superior accuracy on both single- and multi-point mutations and demonstrates strong performance in antibody affinity optimization against H1N1 and HER2, and in predicting the impact of SARS-CoV-2 variants on hACE2 binding. Ablation studies confirm the contribution of each component. These results highlight USP-ddG as a robust and data-efficient framework for modeling mutational effects on PPIs.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Availability and implementation<\/jats:title>\n                    <jats:p>USP-ddG is available at https:\/\/github.com\/ak422\/USP-ddG.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.1093\/bioinformatics\/btag249","type":"journal-article","created":{"date-parts":[[2026,7,7]],"date-time":"2026-07-07T16:29:44Z","timestamp":1783441784000},"source":"Crossref","is-referenced-by-count":0,"title":["USP-ddG: a unified structural paradigm with data efficacy and mixture-of-experts for predicting mutational effects on protein\u2013protein interactions"],"prefix":"10.1093","volume":"42","author":[{"given":"Guanglei","family":"Yu","sequence":"first","affiliation":[{"name":"School of Computer Science and Engineering, Central South University , Changsha 410083,","place":["China"]},{"name":"Hunan Provincial Key Lab on Bioinformatics, Central South University , Changsha 410083,","place":["China"]},{"name":"College of Medical Engineering and Technology, Xinjiang Medical University , Urumqi 830017,","place":["China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Xuehua","family":"Bi","sequence":"additional","affiliation":[{"name":"College of Medical Engineering and Technology, Xinjiang Medical University , Urumqi 830017,","place":["China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-8319-9793","authenticated-orcid":false,"given":"Qichang","family":"Zhao","sequence":"additional","affiliation":[{"name":"School of Computer Science and Engineering, Central South University , Changsha 410083,","place":["China"]},{"name":"Hunan Provincial Key Lab on Bioinformatics, Central South University , Changsha 410083,","place":["China"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-1516-0480","authenticated-orcid":false,"given":"Jianxin","family":"Wang","sequence":"additional","affiliation":[{"name":"School of Computer Science and Engineering, Central South University , Changsha 410083,","place":["China"]},{"name":"Hunan Provincial Key Lab on Bioinformatics, Central South University , Changsha 410083,","place":["China"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"286","published-online":{"date-parts":[[2026,7,7]]},"reference":[{"key":"2026071008225008700_btag249-B1","doi-asserted-by":"crossref","first-page":"493","DOI":"10.1038\/s41586-024-07487-w","article-title":"Accurate structure prediction of biomolecular interactions with AlphaFold 3","volume":"630","author":"Abramson","year":"2024","journal-title":"Nature"},{"key":"2026071008225008700_btag249-B2","doi-asserted-by":"crossref","first-page":"746","DOI":"10.1038\/s41467-022-28313-9","article-title":"Protein sequence design with a learned potential","volume":"13","author":"Anand","year":"2022","journal-title":"Nat Commun"},{"key":"2026071008225008700_btag249-B3","volume-title":"Atkins\u2019 Physical Chemistry","author":"Atkins","year":"2023"},{"key":"2026071008225008700_btag249-B4","first-page":"1352","author":"Bachlechner","year":"2021"},{"key":"2026071008225008700_btag249-B5","doi-asserted-by":"crossref","first-page":"5389","DOI":"10.1021\/acs.jpcb.7b11367","article-title":"Flex ddG: Rosetta ensemble-based estimation of changes in protein\u2013protein binding affinity upon mutation","volume":"122","author":"Barlow","year":"2018","journal-title":"J Phys Chem B"},{"key":"2026071008225008700_btag249-B6","author":"Bushuiev","year":"2024"},{"key":"2026071008225008700_btag249-B7","doi-asserted-by":"crossref","first-page":"7785","DOI":"10.1038\/s41467-024-51563-8","article-title":"Pretrainable geometric graph neural network for antibody affinity maturation","volume":"15","author":"Cai","year":"2024","journal-title":"Nat Commun"},{"key":"2026071008225008700_btag249-B8","author":"Dai","year":"2025"},{"key":"2026071008225008700_btag249-B9","doi-asserted-by":"crossref","first-page":"49","DOI":"10.1126\/science.add2187","article-title":"Robust deep learning based protein sequence design using ProteinMPNN","volume":"378","author":"Dauparas","year":"2022","journal-title":"Science"},{"key":"2026071008225008700_btag249-B10","doi-asserted-by":"crossref","first-page":"4168","DOI":"10.1093\/bioinformatics\/btz184","article-title":"FoldX 5.0: working with RNA, small molecules and a new graphical interface","volume":"35","author":"Delgado","year":"2019","journal-title":"Bioinformatics"},{"key":"2026071008225008700_btag249-B11","volume-title":"Proceedings of the 2023 ICML Workshop on Computational Biology","author":"Dreyer","year":"2023"},{"key":"2026071008225008700_btag249-B12","doi-asserted-by":"crossref","first-page":"1188","DOI":"10.1016\/j.cell.2012.02.022","article-title":"The amyloid state of proteins in human diseases","volume":"148","author":"Eisenberg","year":"2012","journal-title":"Cell"},{"key":"2026071008225008700_btag249-B13","doi-asserted-by":"crossref","first-page":"6501","DOI":"10.1016\/j.cell.2024.10.038","article-title":"Discovery and significance of protein\u2013protein interactions in health and disease","volume":"187","author":"Greenblatt","year":"2024","journal-title":"Cell"},{"key":"2026071008225008700_btag249-B14","doi-asserted-by":"crossref","first-page":"369","DOI":"10.1016\/S0022-2836(02)00442-4","article-title":"Predicting changes in the stability of proteins and protein complexes: a study of more than 1000 mutations","volume":"320","author":"Guerois","year":"2002","journal-title":"J Mol Biol"},{"key":"2026071008225008700_btag249-B15","doi-asserted-by":"crossref","first-page":"441","DOI":"10.1016\/j.jmb.2011.12.062","article-title":"Inhibition of amyloid formation","volume":"421","author":"H\u00e4rd","year":"2012","journal-title":"J Mol Biol"},{"key":"2026071008225008700_btag249-B16","author":"Hu","year":"2022"},{"key":"2026071008225008700_btag249-B17","doi-asserted-by":"crossref","first-page":"462","DOI":"10.1093\/bioinformatics\/bty635","article-title":"SKEMPI 2.0: an updated benchmark of changes in protein\u2013protein binding energy, kinetics and thermodynamics upon mutation","volume":"35","author":"Jankauskaite","year":"2019","journal-title":"Bioinformatics"},{"key":"2026071008225008700_btag249-B18","first-page":"56810","author":"Jiao","year":"2025"},{"key":"2026071008225008700_btag249-B19","author":"Jin","year":"2023"},{"key":"2026071008225008700_btag249-B20","doi-asserted-by":"crossref","first-page":"20120835","DOI":"10.1098\/rsif.2012.0835","article-title":"On the binding affinity of macromolecular interactions: daring to ask why proteins interact","volume":"10","author":"Kastritis","year":"2013","journal-title":"J R Soc Interface"},{"key":"2026071008225008700_btag249-B21","doi-asserted-by":"crossref","first-page":"547","DOI":"10.1038\/nrd2907","article-title":"Cell cycle kinases as therapeutic targets for cancer","volume":"8","author":"Lapenna","year":"2009","journal-title":"Nat Rev Drug Discov"},{"key":"2026071008225008700_btag249-B22","author":"Liu","year":"2024"},{"key":"2026071008225008700_btag249-B23","first-page":"48994","author":"Liu","year":"2023"},{"key":"2026071008225008700_btag249-B24","doi-asserted-by":"crossref","first-page":"e1009284","DOI":"10.1371\/journal.pcbi.1009284","article-title":"Deep geometric representations for modeling effects of mutations on protein\u2013protein binding affinity","volume":"17","author":"Liu","year":"2021","journal-title":"PLoS Comput Biol"},{"key":"2026071008225008700_btag249-B25","author":"Lombardi","year":"2024"},{"key":"2026071008225008700_btag249-B26","author":"Luo","year":"2023"},{"key":"2026071008225008700_btag249-B27","doi-asserted-by":"crossref","first-page":"797","DOI":"10.1038\/nchembio.251","article-title":"Computer-aided design of functional protein interactions","volume":"5","author":"Mandell","year":"2009","journal-title":"Nat Chem Biol"},{"key":"2026071008225008700_btag249-B28","doi-asserted-by":"crossref","first-page":"102370","DOI":"10.1016\/j.sbi.2022.102370","article-title":"Computational design of novel protein\u2013protein interactions\u2014an overview on methodological approaches and applications","volume":"74","author":"Marchand","year":"2022","journal-title":"Curr Opin Struct Biol"},{"key":"2026071008225008700_btag249-B29","doi-asserted-by":"crossref","first-page":"i544","DOI":"10.1093\/bioinformatics\/btad231","article-title":"Deep local analysis deconstructs protein\u2013protein interfaces and accurately estimates binding affinity changes upon mutation","volume":"39","author":"Mohseni Behbahani","year":"2023","journal-title":"Bioinformatics"},{"key":"2026071008225008700_btag249-B30","first-page":"97116","author":"Muennighoff","year":"2025"},{"key":"2026071008225008700_btag249-B31","doi-asserted-by":"crossref","first-page":"btad029","DOI":"10.1093\/bioinformatics\/btad029","article-title":"CATHe: detection of remote homologues for CATH superfamilies using embeddings from protein language models","volume":"39","author":"Nallapareddy","year":"2023","journal-title":"Bioinformatics"},{"key":"2026071008225008700_btag249-B32","doi-asserted-by":"crossref","first-page":"e71393","DOI":"10.7554\/eLife.71393","article-title":"Binding affinity landscapes constrain the evolution of broadly neutralizing anti-influenza antibodies","volume":"10","author":"Phillips","year":"2021","journal-title":"eLife"},{"key":"2026071008225008700_btag249-B33","doi-asserted-by":"crossref","first-page":"e2300838121","DOI":"10.1073\/pnas.2300838121","article-title":"Learning the shape of protein microenvironments with a holographic convolutional neural network","volume":"121","author":"Pun","year":"2024","journal-title":"Proc Natl Acad Sci USA"},{"key":"2026071008225008700_btag249-B34","first-page":"4974","author":"Robinson","year":"2021"},{"key":"2026071008225008700_btag249-B35","doi-asserted-by":"crossref","first-page":"1565","DOI":"10.1126\/science.1203486","article-title":"Cancer immunoediting: integrating immunity\u2019s roles in cancer suppression and promotion","volume":"331","author":"Schreiber","year":"2011","journal-title":"Science"},{"key":"2026071008225008700_btag249-B36","doi-asserted-by":"crossref","first-page":"W382","DOI":"10.1093\/nar\/gki387","article-title":"The FoldX web server: an online force field","volume":"33","author":"Schymkowitz","year":"2005","journal-title":"Nucleic Acids Res"},{"key":"2026071008225008700_btag249-B37","doi-asserted-by":"crossref","first-page":"e2122954119","DOI":"10.1073\/pnas.2122954119","article-title":"Deep learning guided optimization of human antibody against SARS-CoV-2 variants with broad neutralization","volume":"119","author":"Shan","year":"2022","journal-title":"Proc Natl Acad Sci USA"},{"key":"2026071008225008700_btag249-B38","author":"Shanehsazzadeh","year":"2023"},{"key":"2026071008225008700_btag249-B39","author":"Shazeer","year":"2017"},{"key":"2026071008225008700_btag249-B40","first-page":"2597","author":"Soga","year":"2025"},{"key":"2026071008225008700_btag249-B41","doi-asserted-by":"crossref","first-page":"1192","DOI":"10.1038\/s42256-025-01080-0","article-title":"The importance of negative training data for robust antibody binding prediction","volume":"7","author":"Ta","year":"2025","journal-title":"Nat Mach Intell"},{"key":"2026071008225008700_btag249-B42","doi-asserted-by":"crossref","first-page":"1206","DOI":"10.1038\/s42256-025-01089-5","article-title":"Training data composition determines machine learning generalization and biological rule discovery","volume":"7","author":"Ursu","year":"2025","journal-title":"Nat Mach Intell"},{"key":"2026071008225008700_btag249-B43","doi-asserted-by":"crossref","first-page":"16649","DOI":"10.1021\/acs.jmedchem.4c01459","article-title":"Inhibition of the EphA2-Sam\/Ship2-Sam association through peptide ligands: studying the combined effect of charge and aromatic character","volume":"67","author":"Vincenzi","year":"2024","journal-title":"J Med Chem"},{"key":"2026071008225008700_btag249-B44","doi-asserted-by":"crossref","first-page":"2007","DOI":"10.1038\/s41591-023-02483-5","article-title":"Deep-learning-enabled protein\u2013protein interaction analysis for prediction of SARS-CoV-2 infectivity and variant evolution","volume":"29","author":"Wang","year":"2023","journal-title":"Nat Med"},{"key":"2026071008225008700_btag249-B45","author":"Wu","year":"2024"},{"key":"2026071008225008700_btag249-B46","doi-asserted-by":"crossref","first-page":"323","DOI":"10.1186\/s12859-020-03646-8","article-title":"Graph-based prediction of protein\u2013protein interactions with attributed signed graph embedding","volume":"21","author":"Yang","year":"2020","journal-title":"BMC Bioinformatics"},{"key":"2026071008225008700_btag249-B47","doi-asserted-by":"crossref","first-page":"i362","DOI":"10.1093\/bioinformatics\/btaf228","article-title":"CATH-ddG: towards robust mutation effect prediction on protein\u2013protein interactions out of CATH homologous superfamily","volume":"41","author":"Yu","year":"2025","journal-title":"Bioinformatics"},{"key":"2026071008225008700_btag249-B48","first-page":"23219","author":"Yu","year":"2024"},{"key":"2026071008225008700_btag249-B49","author":"Zhao","year":"2025"}],"container-title":["Bioinformatics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/academic.oup.com\/bioinformatics\/article-pdf\/42\/Supplement_1\/btag249\/68709168\/btag249.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/bioinformatics\/article-pdf\/42\/Supplement_1\/btag249\/68709168\/btag249.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,7,10]],"date-time":"2026-07-10T12:23:01Z","timestamp":1783686181000},"score":1,"resource":{"primary":{"URL":"https:\/\/academic.oup.com\/bioinformatics\/article\/doi\/10.1093\/bioinformatics\/btag249\/8726290"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,7,1]]},"references-count":49,"journal-issue":{"issue":"Supplement_1","published-print":{"date-parts":[[2026,7,1]]}},"URL":"https:\/\/doi.org\/10.1093\/bioinformatics\/btag249","relation":{},"ISSN":["1367-4803","1367-4811"],"issn-type":[{"value":"1367-4803","type":"print"},{"value":"1367-4811","type":"electronic"}],"subject":[],"published-other":{"date-parts":[[2026,7]]},"published":{"date-parts":[[2026,7,1]]},"article-number":"btag249"}}