{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"institution":[{"name":"bioRxiv"}],"indexed":{"date-parts":[[2026,7,30]],"date-time":"2026-07-30T13:17:50Z","timestamp":1785417470149,"version":"3.56.0"},"posted":{"date-parts":[[2025,11,14]]},"group-title":"Bioinformatics","reference-count":43,"publisher":"openRxiv","license":[{"start":{"date-parts":[[2025,11,14]],"date-time":"2025-11-14T00:00:00Z","timestamp":1763078400000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by-nc-nd\/4.0\/"}],"funder":[{"award":["P01CA196569"],"award-info":[{"award-number":["P01CA196569"]}],"id":[{"id":"https:\/\/ror.org\/01cwqze88","id-type":"ROR","asserted-by":"publisher"}]},{"award":["U01CA261339"],"award-info":[{"award-number":["U01CA261339"]}],"id":[{"id":"https:\/\/ror.org\/01cwqze88","id-type":"ROR","asserted-by":"publisher"}]},{"award":["5P30CA 014089-47"],"award-info":[{"award-number":["5P30CA 014089-47"]}],"id":[{"id":"https:\/\/ror.org\/01cwqze88","id-type":"ROR","asserted-by":"publisher"}]},{"award":["P30ES007048"],"award-info":[{"award-number":["P30ES007048"]}],"id":[{"id":"https:\/\/ror.org\/01cwqze88","id-type":"ROR","asserted-by":"publisher"}]},{"award":["U01CA164973"],"award-info":[{"award-number":["U01CA164973"]}],"id":[{"id":"https:\/\/ror.org\/01cwqze88","id-type":"ROR","asserted-by":"publisher"}]},{"award":["P42ES036506"],"award-info":[{"award-number":["P42ES036506"]}],"id":[{"id":"https:\/\/ror.org\/01cwqze88","id-type":"ROR","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"accepted":{"date-parts":[[2026,4,30]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                <jats:sec>\n                  <jats:title>Motivation<\/jats:title>\n                  <jats:p>High-dimensional omics data are typically measured on limited sample sizes, which challenges model-based clustering methods such as Gaussian mixture models, often leading to instability and poor generalization under complex mixture structures. To address these limitations, we developed Praxis-BGM, a natural-gradient variational inference framework for Gaussian mixture models that enables semi-supervised transfer learning by incorporating an informative prior Gaussian mixture model derived from large-scale reference data with robust cluster structures. This prior can encode cluster-specific means, covariance structures, and structural connectivity patterns, and is updated using the target data with variational inference to improve clustering in small-sample settings.<\/jats:p>\n                <\/jats:sec>\n                <jats:sec>\n                  <jats:title>Results<\/jats:title>\n                  <jats:p>We derived natural-gradient updates for standard parameters and assess feature-level contributions to posterior clustering via Bayes Factors. Implemented in Python library JAX for accelerator-oriented computation, Praxis-BGM is computationally efficient and scalable. Across extensive simulations and two real-world applications\u2014breast cancer bulk transcriptomics for subtype recovery and single-cell transcriptomics for cross-platform label transfer\u2014Praxis-BGM improves posterior clustering performance, stability, and biological interpretability, even when priors are partially mismatched.<\/jats:p>\n                <\/jats:sec>\n                <jats:sec>\n                  <jats:title>Availability and Implementation<\/jats:title>\n                  <jats:p>\n                    Praxis-BGM is freely available at\n                    <jats:ext-link xmlns:xlink=\"http:\/\/www.w3.org\/1999\/xlink\" ext-link-type=\"uri\" xlink:href=\"https:\/\/github.com\/ContiLab-usc\/Praxis-BGM\">https:\/\/github.com\/ContiLab-usc\/Praxis-BGM<\/jats:ext-link>\n                    , and an archival version is available on Zenodo at\n                    <jats:ext-link xmlns:xlink=\"http:\/\/www.w3.org\/1999\/xlink\" ext-link-type=\"uri\" xlink:href=\"https:\/\/doi.org\/10.5281\/zenodo.19657680\">https:\/\/doi.org\/10.5281\/zenodo.19657680<\/jats:ext-link>\n                    .\n                  <\/jats:p>\n                <\/jats:sec>\n                <jats:sec>\n                  <jats:title>Contact<\/jats:title>\n                  <jats:p>\n                    <jats:email>qiranjia@usc.edu<\/jats:email>\n                  <\/jats:p>\n                <\/jats:sec>\n                <jats:sec>\n                  <jats:title>Supplementary Information<\/jats:title>\n                  <jats:p>Supplementary materials are available with the manuscript submission.<\/jats:p>\n                <\/jats:sec>","DOI":"10.1101\/2025.11.13.688299","type":"posted-content","created":{"date-parts":[[2025,11,14]],"date-time":"2025-11-14T18:55:18Z","timestamp":1763146518000},"source":"Crossref","is-referenced-by-count":0,"title":["Praxis-BGM: Clustering of Omics Data Using Semi-Supervised Transfer Learning for Gaussian Mixture Models via Natural-Gradient Variational Inference"],"prefix":"10.64898","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-0790-5967","authenticated-orcid":false,"given":"Qiran","family":"Jia","sequence":"first","affiliation":[{"name":"Division of Biostatistics and Health Data Science, Department of Population and Public Health Sciences, Keck School of Medicine, University of Southern California"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6615-0472","authenticated-orcid":false,"given":"Jesse A.","family":"Goodrich","sequence":"additional","affiliation":[{"name":"Division of Environmental Health, Department of Population and Public Health Sciences, Keck School of Medicine, University of Southern California"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-2941-7833","authenticated-orcid":false,"given":"David V.","family":"Conti","sequence":"additional","affiliation":[{"name":"Division of Biostatistics and Health Data Science, Department of Population and Public Health Sciences, Keck School of Medicine, University of Southern California"},{"name":"Department of Biostatistics and Informatics, Colorado School of Public Health"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"54368","reference":[{"key":"2026050400300634000_2025.11.13.688299v3.1","doi-asserted-by":"publisher","DOI":"10.1162\/089976698300017746"},{"key":"2026050400300634000_2025.11.13.688299v3.2","doi-asserted-by":"publisher","DOI":"10.1038\/75556"},{"key":"2026050400300634000_2025.11.13.688299v3.3","doi-asserted-by":"publisher","DOI":"10.7150\/thno.11543"},{"key":"2026050400300634000_2025.11.13.688299v3.4","doi-asserted-by":"publisher","DOI":"10.1080\/01621459.2017.1285773"},{"key":"2026050400300634000_2025.11.13.688299v3.5","unstructured":"James Bradbury , Roy Frostig , Peter Hawkins , Matthew James Johnson , Yash Katariya , Chris Leary , Dougal Maclaurin , George Necula , Adam Paszke , Jake VanderPlas , Skye Wanderman-Milne , and Qiao Zhang . Jax: composable transformations of python+numpy programs, 2018."},{"key":"2026050400300634000_2025.11.13.688299v3.6","doi-asserted-by":"publisher","DOI":"10.1158\/2159-8290.CD-12-0095"},{"key":"2026050400300634000_2025.11.13.688299v3.7","doi-asserted-by":"crossref","unstructured":"Tianqi Chen and Carlos Guestrin . Xgboost: A scalable tree boosting system. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD \u201816), pages 785\u2013794, New York, NY, USA, 2016. Association for Computing Machinery.","DOI":"10.1145\/2939672.2939785"},{"issue":"D1","key":"2026050400300634000_2025.11.13.688299v3.8","doi-asserted-by":"crossref","first-page":"D1779","DOI":"10.1093\/nar\/gkaf1292","article-title":"The gene ontology knowledgebase in 2026","volume":"54","author":"The Gene Ontology Consortium","year":"2026","journal-title":"Nucleic Acids Research"},{"key":"2026050400300634000_2025.11.13.688299v3.9","doi-asserted-by":"publisher","DOI":"10.1038\/nature10983"},{"key":"2026050400300634000_2025.11.13.688299v3.10","doi-asserted-by":"publisher","DOI":"10.1158\/0008-5472.Can-23-0816"},{"key":"2026050400300634000_2025.11.13.688299v3.11","doi-asserted-by":"publisher","DOI":"10.1093\/bioinformatics\/btac757"},{"key":"2026050400300634000_2025.11.13.688299v3.12","doi-asserted-by":"publisher","DOI":"10.1198\/016214502760047131"},{"key":"2026050400300634000_2025.11.13.688299v3.13","doi-asserted-by":"publisher","DOI":"10.1126\/scisignal.2004088"},{"key":"2026050400300634000_2025.11.13.688299v3.14","doi-asserted-by":"publisher","DOI":"10.1016\/j.envint.2024.108930"},{"key":"2026050400300634000_2025.11.13.688299v3.15","doi-asserted-by":"publisher","DOI":"10.1016\/J.CELL.2021.04.048"},{"issue":"8","key":"2026050400300634000_2025.11.13.688299v3.16","doi-asserted-by":"crossref","first-page":"2283","DOI":"10.1038\/s41596-024-00991-3","article-title":"Scanorama: integrating large and diverse single-cell transcriptomic datasets","volume":"19","year":"2024","journal-title":"Nature Protocols"},{"key":"2026050400300634000_2025.11.13.688299v3.17","unstructured":"Harold Jeffreys . Theory of Probability. Oxford University Press, 3rd edition, 1961."},{"key":"2026050400300634000_2025.11.13.688299v3.18","doi-asserted-by":"crossref","unstructured":"Mohammad Emtiyaz Khan and Didrik Nielsen . Fast yet simple natural-gradient descent for variational inference in complex models. In 2018 International Symposium on Information Theory and Its Applications (ISITA), pages 31\u201335. IEEE, 2018.","DOI":"10.23919\/ISITA.2018.8664326"},{"issue":"1","key":"2026050400300634000_2025.11.13.688299v3.19","doi-asserted-by":"crossref","first-page":"149","DOI":"10.1111\/rssb.12479","article-title":"Transfer learning for high-dimensional linear regression: Prediction, estimation and minimax optimality","volume":"84","year":"2022","journal-title":"Journal of the Royal Statistical Society Series B: Statistical Methodology"},{"issue":"6","key":"2026050400300634000_2025.11.13.688299v3.20","doi-asserted-by":"crossref","first-page":"417","DOI":"10.1016\/j.cels.2015.12.004","article-title":"The molecular signatures database (msigdb) hallmark gene set collection","volume":"1","year":"2015","journal-title":"Cell Systems"},{"key":"2026050400300634000_2025.11.13.688299v3.21","unstructured":"Wu Lin , Mohammad Emtiyaz Khan , and Mark Schmidt . Fast and simple natural-gradient variational inference with mixture of exponential-family approximations. In Proceedings of the 36th International Conference on Machine Learning, volume 97 of Proceedings of Machine Learning Research, pages 3992\u20134002. PMLR, 2019."},{"key":"2026050400300634000_2025.11.13.688299v3.22","doi-asserted-by":"crossref","first-page":"1053","DOI":"10.1038\/s41592-018-0229-2","article-title":"Deep generative modeling for single-cell transcriptomics","volume":"15","year":"2018","journal-title":"Nature Methods"},{"issue":"1","key":"2026050400300634000_2025.11.13.688299v3.23","doi-asserted-by":"crossref","first-page":"41","DOI":"10.1038\/s41592-021-01336-8","article-title":"Benchmarking atlas-level data integration in single-cell genomics","volume":"19","year":"2022","journal-title":"Nature Methods"},{"key":"2026050400300634000_2025.11.13.688299v3.24","article-title":"Natural gradient variational inference with gaussian mixture models","year":"2021","journal-title":"arXiv preprint"},{"key":"2026050400300634000_2025.11.13.688299v3.25","doi-asserted-by":"publisher","DOI":"10.1073\/pnas.1208949110"},{"key":"2026050400300634000_2025.11.13.688299v3.26","doi-asserted-by":"publisher","DOI":"10.1038\/nature11412"},{"issue":"10","key":"2026050400300634000_2025.11.13.688299v3.27","doi-asserted-by":"crossref","first-page":"1345","DOI":"10.1109\/TKDE.2009.191","article-title":"A survey on transfer learning","volume":"22","year":"2010","journal-title":"IEEE Transactions on Knowledge and Data Engineering"},{"key":"2026050400300634000_2025.11.13.688299v3.28","doi-asserted-by":"publisher","DOI":"10.1200\/JCO.2008.18.1370"},{"key":"2026050400300634000_2025.11.13.688299v3.29","doi-asserted-by":"crossref","first-page":"961","DOI":"10.1016\/j.csbj.2021.01.015","article-title":"Automated methods for cell type annotation on scrna-seq data","volume":"19","year":"2021","journal-title":"Computational and Structural Biotechnology Journal"},{"key":"2026050400300634000_2025.11.13.688299v3.30","first-page":"2825","article-title":"Scikit-learn: Machine learning in python","volume":"12","year":"2011","journal-title":"Journal of Machine Learning Research"},{"key":"2026050400300634000_2025.11.13.688299v3.31","doi-asserted-by":"publisher","DOI":"10.1038\/35021093"},{"key":"2026050400300634000_2025.11.13.688299v3.32","doi-asserted-by":"publisher","DOI":"10.1200\/CCI.19.00119"},{"key":"2026050400300634000_2025.11.13.688299v3.33","doi-asserted-by":"publisher","DOI":"10.1093\/nar\/gky889"},{"issue":"3","key":"2026050400300634000_2025.11.13.688299v3.34","first-page":"477","article-title":"Bayesian transfer learning","volume":"40","year":"2025","journal-title":"Statistical Science"},{"key":"2026050400300634000_2025.11.13.688299v3.35","doi-asserted-by":"publisher","DOI":"10.1186\/1479-7364-4-1-21"},{"key":"2026050400300634000_2025.11.13.688299v3.36","article-title":"Robust unsupervised multi-task and transfer learning on gaussian mixture models","year":"2024","journal-title":"arXiv preprint"},{"issue":"2247","key":"2026050400300634000_2025.11.13.688299v3.37","doi-asserted-by":"crossref","first-page":"20220149","DOI":"10.1098\/rsta.2022.0149","article-title":"Bayesian cluster analysis","volume":"381","year":"2023","journal-title":"Philosophical Transactions of the Royal Society A"},{"issue":"11","key":"2026050400300634000_2025.11.13.688299v3.38","doi-asserted-by":"crossref","first-page":"7058","DOI":"10.1109\/TCYB.2022.3177242","article-title":"Transfer-learning-based gaussian mixture model for distributed clustering","volume":"53","year":"2023","journal-title":"IEEE Transactions on Cybernetics"},{"key":"2026050400300634000_2025.11.13.688299v3.39","article-title":"Formal bayesian transfer learning via the total risk prior","year":"2025","journal-title":"arXiv preprint"},{"key":"2026050400300634000_2025.11.13.688299v3.40","doi-asserted-by":"publisher","DOI":"10.15252\/msb.20209620"},{"key":"2026050400300634000_2025.11.13.688299v3.41","doi-asserted-by":"crossref","first-page":"116615","DOI":"10.1016\/j.biopha.2024.116615","article-title":"Targeting the crosstalk between estrogen receptors and membrane growth factor receptors in breast cancer treatment: Advances and opportunities","volume":"175","year":"2024","journal-title":"Biomedicine & Pharmacotherapy"},{"issue":"3","key":"2026050400300634000_2025.11.13.688299v3.42","doi-asserted-by":"crossref","first-page":"qaa078","DOI":"10.1093\/nargab\/lqaa078","article-title":"Combat-seq: batch effect adjustment for rna-seq count data","volume":"2","year":"2020","journal-title":"NAR Genomics and Bioinformatics"},{"issue":"2","key":"2026050400300634000_2025.11.13.688299v3.43","doi-asserted-by":"crossref","first-page":"4","DOI":"10.32614\/RJ-2024-012","article-title":"Lucidus: An r package for implementing latent unknown clustering by integrating multi-omics data (lucid) with phenotypic traits","volume":"16","year":"2025","journal-title":"The R Journal"}],"container-title":[],"original-title":[],"link":[{"URL":"https:\/\/syndication.highwire.org\/content\/doi\/10.1101\/2025.11.13.688299","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,5,4]],"date-time":"2026-05-04T07:30:19Z","timestamp":1777879819000},"score":1,"resource":{"primary":{"URL":"http:\/\/biorxiv.org\/lookup\/doi\/10.1101\/2025.11.13.688299"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,11,14]]},"references-count":43,"URL":"https:\/\/doi.org\/10.1101\/2025.11.13.688299","relation":{},"subject":[],"published":{"date-parts":[[2025,11,14]]},"subtype":"preprint"}}