{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,13]],"date-time":"2026-07-13T16:04:35Z","timestamp":1783958675990,"version":"3.55.0"},"reference-count":39,"publisher":"Oxford University Press (OUP)","issue":"7","license":[{"start":{"date-parts":[[2026,6,13]],"date-time":"2026-06-13T00:00:00Z","timestamp":1781308800000},"content-version":"vor","delay-in-days":1,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/100020595","name":"National Science and Technology Council","doi-asserted-by":"publisher","award":["NSTC114-2221-E-038-015"],"award-info":[{"award-number":["NSTC114-2221-E-038-015"]}],"id":[{"id":"10.13039\/100020595","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100020595","name":"National Science and Technology Council","doi-asserted-by":"publisher","award":["NSTC115-2221-E-038-012-MY3"],"award-info":[{"award-number":["NSTC115-2221-E-038-012-MY3"]}],"id":[{"id":"10.13039\/100020595","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2026,7,2]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:sec>\n                    <jats:title>Motivation<\/jats:title>\n                    <jats:p>Foundation models such as scGPT have demonstrated strong potential for single-cell multi-omics integration; however, their downstream performance is highly sensitive to hyperparameter selection. Manual fine-tuning remains computationally expensive, dataset-dependent, and often irreproducible. Despite the increasing adoption of foundation models in single-cell analysis, systematic strategies for robust hyperparameter optimization remain underexplored.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>We developed a Bayesian optimization framework based on Tree-structured Parzen Estimators (TPE) for automated fine-tuning of scGPT and evaluated its performance on two benchmark bone marrow mononuclear cell (BMMC) multi-omics datasets, including CITE-seq and GSE194122 datasets. Across datasets, Bayesian optimization consistently improved biological conservation and batch integration metrics compared with default scGPT configurations. On the original BMMC benchmark, optimization improved AvgBIO from 0.59 to 0.67 and PCR from 0.33 to 0.52. On the GSE194122 dataset, the default configuration exhibited unstable convergence and weak biological preservation (AvgBIO = 0.19; ARI = 0.007), whereas Bayesian optimization substantially improved integration performance (AvgBIO = 0.60; ARI = 0.63) while reducing validation loss from 137 to 47.1. These findings demonstrate substantial dataset-specific sensitivity of scGPT fine-tuning and highlight the importance of automated optimization for stable deployment across heterogeneous multi-omics datasets. Our study demonstrates that Bayesian optimization provides an effective and reproducible strategy for stabilizing scGPT fine-tuning across diverse single-cell multi-omics datasets. Rather than introducing a new integration architecture, this work emphasizes the importance of systematic optimization for improving robustness and reproducibility of foundation-model applications in computational biology.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Availability and implementation<\/jats:title>\n                    <jats:p>Our model and dataset are freely available at: https:\/\/github.com\/daren642\/scGPT_multiomic_tuning.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.1093\/bioinformatics\/btag374","type":"journal-article","created":{"date-parts":[[2026,6,10]],"date-time":"2026-06-10T11:51:55Z","timestamp":1781092315000},"source":"Crossref","is-referenced-by-count":0,"title":["Bayesian hyperparameter optimization improves scGPT fine-tuning for single-cell multi-omics integration"],"prefix":"10.1093","volume":"42","author":[{"given":"Darren Yu Jun","family":"Tay","sequence":"first","affiliation":[{"name":"Science Research Programme, Catholic Junior College , Singapore 297822,","place":["Singapore"]},{"name":"AIBioMed Lab, Taipei Medical University , Taipei 110,","place":["Taiwan"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-4896-7926","authenticated-orcid":false,"given":"Nguyen Quoc Khanh","family":"Le","sequence":"additional","affiliation":[{"name":"AIBioMed Lab, Taipei Medical University , Taipei 110,","place":["Taiwan"]},{"name":"In-Service Master Program in Artificial Intelligence in Medicine, College of Medicine, Taipei Medical University , Taipei 110,","place":["Taiwan"]},{"name":"Translational Imaging Research Center, Taipei Medical University Hospital , Taipei 110,","place":["Taiwan"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Matthew Chin Heng","family":"Chua","sequence":"additional","affiliation":[{"name":"Department of Biomedical Informatics, Yong Loo Lin School of Medicine, National University of Singapore , Singapore 119615,","place":["Singapore"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"286","published-online":{"date-parts":[[2026,6,12]]},"reference":[{"key":"2026071311072128400_btag374-B1","doi-asserted-by":"crossref","first-page":"3006","DOI":"10.3390\/math12193006","article-title":"TPE-optimized DNN with attention mechanism for prediction of tower crane payload moving conditions","volume":"12","author":"Akber","year":"2024","journal-title":"Mathematics"},{"key":"2026071311072128400_btag374-B2","doi-asserted-by":"crossref","first-page":"85","DOI":"10.1016\/j.coisb.2017.07.004","article-title":"Single cells make big data: new challenges and opportunities in transcriptomics","volume":"4","author":"Angerer","year":"2017","journal-title":"Curr Opin Syst Biol"},{"key":"2026071311072128400_btag374-B3","first-page":"2546","article-title":"Algorithms for hyper-parameter optimization","volume":"24","author":"Bergstra","year":"2011","journal-title":"Adv Neural Inf Process Syst"},{"key":"2026071311072128400_btag374-B4","first-page":"281","article-title":"Random search for hyper-parameter optimization","volume":"13","author":"Bergstra","year":"2012","journal-title":"J Mach Learn Res"},{"key":"2026071311072128400_btag374-B5","doi-asserted-by":"crossref","first-page":"014008","DOI":"10.1088\/1749-4699\/8\/1\/014008","article-title":"Hyperopt: a python library for model selection and hyperparameter optimization","volume":"8","author":"Bergstra","year":"2015","journal-title":"Comput Sci Discov"},{"key":"2026071311072128400_btag374-B6","first-page":"548","author":"Boudiaf","year":"2020"},{"key":"2026071311072128400_btag374-B7","author":"Bousquet","year":"2011"},{"key":"2026071311072128400_btag374-B8","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1038\/s43018-025-00917-2","article-title":"Hallmarks of artificial intelligence contributions to precision oncology","volume":"6","author":"Chang","year":"2025","journal-title":"Nat Cancer"},{"key":"2026071311072128400_btag374-B9","doi-asserted-by":"crossref","first-page":"309","DOI":"10.1016\/j.ijmst.2021.08.004","article-title":"Machine learning-based classification of rock discontinuity trace: smote oversampling integrated with GBT ensemble learning","volume":"32","author":"Chen","year":"2022","journal-title":"Int J Min Sci Technol"},{"key":"2026071311072128400_btag374-B10","first-page":"1879","author":"Claret","year":"2024"},{"key":"2026071311072128400_btag374-B11","doi-asserted-by":"crossref","first-page":"1470","DOI":"10.1038\/s41592-024-02201-0","article-title":"SCGPT: toward building a foundation model for single-cell multi-omics using generative AI","volume":"21","author":"Cui","year":"2024","journal-title":"Nat Methods"},{"key":"2026071311072128400_btag374-B12","first-page":"577","author":"Elder","year":"1992"},{"key":"2026071311072128400_btag374-B13","first-page":"1","author":"Fang","year":"2024"},{"key":"2026071311072128400_btag374-B14","doi-asserted-by":"crossref","first-page":"119700","DOI":"10.1016\/j.renene.2023.119700","article-title":"Advanced hyperparameter optimization of deep learning models for wind power prediction","volume":"221","author":"Hanifi","year":"2024","journal-title":"Renewable Energy"},{"key":"2026071311072128400_btag374-B15","doi-asserted-by":"crossref","first-page":"522","DOI":"10.1016\/j.cels.2021.05.016","article-title":"Machine learning for perturbational single-cell omics","volume":"12","author":"Ji","year":"2021","journal-title":"Cell Syst"},{"key":"2026071311072128400_btag374-B16","first-page":"257","author":"Keogh","year":"2011"},{"key":"2026071311072128400_btag374-B17","doi-asserted-by":"crossref","first-page":"4299","DOI":"10.1038\/s41598-024-54212-8","article-title":"A precise model for skin cancer diagnosis using hybrid u-net and improved mobilenet-v3 with hyperparameters optimization","volume":"14","author":"Kumar Lilhore","year":"2024","journal-title":"Sci Rep"},{"key":"2026071311072128400_btag374-B18","doi-asserted-by":"crossref","first-page":"97","DOI":"10.1115\/1.3653121","article-title":"A new method of locating the maximum point of an arbitrary multipeak curve in the presence of noise","volume":"86","author":"Kushner","year":"1964","journal-title":"J Basic Eng"},{"key":"2026071311072128400_btag374-B19","doi-asserted-by":"crossref","first-page":"236","DOI":"10.1186\/s12911-024-02631-y","article-title":"Optimizing protein sequence classification: integrating deep learning models with Bayesian optimization for enhanced biological analysis","volume":"24","author":"Lilhore","year":"2024","journal-title":"BMC Med Inf Dec Mak"},{"key":"2026071311072128400_btag374-B20","doi-asserted-by":"crossref","first-page":"655536","DOI":"10.3389\/fgene.2021.655536","article-title":"Machine intelligence in single-cell data analysis: advances and new challenges","volume":"12","author":"Liu","year":"2021","journal-title":"Front Genet"},{"key":"2026071311072128400_btag374-B21","author":"Luecken","year":"2021"},{"key":"2026071311072128400_btag374-B22","first-page":"23803","author":"Mao","year":"2023"},{"key":"2026071311072128400_btag374-B23","first-page":"400","author":"Mo\u010dkus","year":"1974"},{"key":"2026071311072128400_btag374-B24","doi-asserted-by":"crossref","first-page":"27","DOI":"10.1016\/j.coisb.2021.02.002","article-title":"Algorithmic advances in machine learning for single-cell expression analysis","volume":"25","author":"Oller-Moreno","year":"2021","journal-title":"Curr Opin Syst Biol"},{"key":"2026071311072128400_btag374-B25","author":"Owen","year":"2022"},{"key":"2026071311072128400_btag374-B26","doi-asserted-by":"crossref","first-page":"1209","DOI":"10.1613\/jair.1.13188","article-title":"Multiobjective tree-structured Parzen estimator","volume":"73","author":"Ozaki","year":"2022","journal-title":"J Artif Intell Res"},{"key":"2026071311072128400_btag374-B27","doi-asserted-by":"crossref","first-page":"D886","DOI":"10.1093\/nar\/gkae1142","article-title":"CZ cellxgene discover: a single-cell data platform for scalable exploration, analysis and modeling of aggregated data","volume":"53","author":"Program","year":"2025","journal-title":"Nucleic Acids Res"},{"key":"2026071311072128400_btag374-B28","doi-asserted-by":"crossref","first-page":"1196","DOI":"10.1093\/bib\/bbz062","article-title":"Clustering and classification methods for single-cell RNA-sequencing data","volume":"21","author":"Qi","year":"2020","journal-title":"Brief Bioinf"},{"key":"2026071311072128400_btag374-B29","doi-asserted-by":"crossref","first-page":"bbaa216","DOI":"10.1093\/bib\/bbaa216","article-title":"A spectral clustering with self-weighted multiple kernel learning method for single-cell RNA-seq data","volume":"22","author":"Qi","year":"2021","journal-title":"Brief Bioinf"},{"key":"2026071311072128400_btag374-B30","doi-asserted-by":"crossref","first-page":"e27041","DOI":"10.7554\/eLife.27041","article-title":"The human cell atlas","volume":"6","author":"Regev","year":"2017","journal-title":"elife"},{"key":"2026071311072128400_btag374-B31","doi-asserted-by":"crossref","first-page":"e202451234","DOI":"10.1002\/eji.202451234","article-title":"Towards the next generation of data-driven therapeutics using spatially resolved single-cell technologies and generative AI","volume":"55","author":"Rodov","year":"2025","journal-title":"Eur J Immunol"},{"key":"2026071311072128400_btag374-B32","doi-asserted-by":"crossref","first-page":"148","DOI":"10.1109\/JPROC.2015.2494218","article-title":"Taking the human out of the loop: a review of Bayesian optimization","volume":"104","author":"Shahriari","year":"2015","journal-title":"Proc IEEE"},{"key":"2026071311072128400_btag374-B33","first-page":"20459","author":"Stanton","year":"2022"},{"key":"2026071311072128400_btag374-B34","doi-asserted-by":"crossref","first-page":"965","DOI":"10.1109\/21.23094","article-title":"A global search method for optimizing nonlinear systems","volume":"18","author":"Stuckman","year":"2002","journal-title":"IEEE Trans Syst Man Cybern"},{"key":"2026071311072128400_btag374-B35","doi-asserted-by":"crossref","first-page":"A029","DOI":"10.1158\/1538-7445.PANCREATIC24-A029","article-title":"Abstract a029: evaluating and interpreting SCGPT: a foundation model for single-cell biology in real-world cancer clinical trial data","volume":"84","author":"Tan","year":"2024","journal-title":"Cancer Res"},{"key":"2026071311072128400_btag374-B36","first-page":"5998","article-title":"Attention is all you need","volume":"30","author":"Vaswani","year":"2017","journal-title":"Adv Neural Inf Process Syst"},{"key":"2026071311072128400_btag374-B37","first-page":"1","article-title":"Recent advances in Bayesian optimization","volume":"55","author":"Wang","year":"2023","journal-title":"ACM Comput Surv"},{"key":"2026071311072128400_btag374-B38","doi-asserted-by":"crossref","first-page":"102","DOI":"10.1186\/s40104-024-01063-y","article-title":"Advances in single-cell transcriptomics in animal research","volume":"15","author":"Yan","year":"2024","journal-title":"J Animal Sci Biotechnol"},{"key":"2026071311072128400_btag374-B39","doi-asserted-by":"crossref","first-page":"104914","DOI":"10.1016\/j.ijrmms.2021.104914","article-title":"Multi-source data driven method for assessing the rock mass quality of a NATM tunnel face via hybrid ensemble learning models","volume":"147","author":"Zhou","year":"2021","journal-title":"Int J Rock Mech Min Sci"}],"container-title":["Bioinformatics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/academic.oup.com\/bioinformatics\/advance-article-pdf\/doi\/10.1093\/bioinformatics\/btag374\/68520345\/btag374.pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/bioinformatics\/article-pdf\/42\/7\/btag374\/68520345\/btag374.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/bioinformatics\/article-pdf\/42\/7\/btag374\/68520345\/btag374.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,7,13]],"date-time":"2026-07-13T15:07:32Z","timestamp":1783955252000},"score":1,"resource":{"primary":{"URL":"https:\/\/academic.oup.com\/bioinformatics\/article\/doi\/10.1093\/bioinformatics\/btag374\/8707330"}},"subtitle":[],"editor":[{"given":"Pier Luigi","family":"Martelli","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"editor"}]}],"short-title":[],"issued":{"date-parts":[[2026,6,12]]},"references-count":39,"journal-issue":{"issue":"7","published-print":{"date-parts":[[2026,7,2]]}},"URL":"https:\/\/doi.org\/10.1093\/bioinformatics\/btag374","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":[[2026,6,12]]},"article-number":"btag374"}}