Research BriefingPublished: 10 October 2025Nature Methods (2025)Cite this articleSubjectsComputational modelsMachine learningSoftwareEpiAgent, a transformer-based foundation model pretrained on approximately 5 million cells and over 35 billion tokens, has advanced single-cell epigenomics by encoding chromatin accessibility as ‘cell sentences’. Benefiting from this framework, EpiAgent achieved state-of-the-art performance in typical downstream tasks and enabled perturbation response prediction and in silico chromatin region knockouts.This is a preview of subscription content, access via your institutionAccess optionsAccess Nature and 54 other Nature Portfolio journalsGet Nature+, our best-value online-access subscription27,99 € / 30 dayscancel any timeLearn moreSubscribe to this journalReceive 12 print issues and online access269,00 € per yearonly 22,42 € per issueLearn moreBuy this articlePurchase on SpringerLinkInstant access to full article PDFBuy nowPrices may be subject to local taxes which are calculated during checkoutFig. 1: The pretraining data and model architecture of EpiAgent.ReferencesStricker, S. H., Köferle, A. & Beck, S. From profiles to function in epigenomics. Nat. Rev. Genet. 18, 51–66 (2017). A review article that provides an in-depth overview of epigenomics, from profiling epigenomic features to understanding their functional implications.Article CAS PubMed Google Scholar Minnoye, L. et al. Chromatin accessibility profiling methods. Nat. Rev. Methods Primers 1, 10 (2021). A review article that provides a comprehensive overview of chromatin accessibility sequencing methods, including workflows for scATAC-seq and their research importance.Article CAS PubMed PubMed Central Google Scholar Stuart, T., Srivastava, A., Madad, S., Lareau, C. A. & Satija, R. Single-cell chromatin state analysis with Signac. Nat. Methods 18, 1333–1341 (2021). This paper introduces Signac, a representative and comprehensive pipeline for the analysis of single-cell epigenomic data.Article CAS PubMed PubMed Central Google Scholar Yuan, H. & Kelley, D. R. scBasset: sequence-based modeling of single-cell ATAC-seq using convolutional neural networks. Nat. Methods 19, 1088–1096 (2022). This study presents scBasset, a representative computational approach for scATAC-seq data analysis.Article CAS PubMed Google Scholar Gao, S. et al. Empowering biomedical discovery with AI agents. Cell 187, 6125–6151 (2024). A review article that surveys the applications of AI agents in biomedicine, discussing their types, current developments, associated challenges and future prospects.Article CAS PubMed Google Scholar Download referencesAdditional informationPublisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.This is a summary of: Chen, X. et al. EpiAgent: foundation model for single-cell epigenomics. Nat. Methods https://doi.org/10.1038/s41592-025-02822-z (2025).Rights and permissionsReprints and permissionsAbout this article