Manual analysis of single-molecule time traces is slow and subjective. Now, a transformer-based foundation model — META-SiM —automates key analysis tasks across diverse datasets and enables rapid, systematic discovery of subtle single-molecule behaviors. Application of this approach reveals a previously undetected pre-mRNA splicing intermediate, highlighting its potential to streamline biological discovery.
This is a preview of subscription content, access via your institution
Access options
Access Nature and 54 other Nature Portfolio journals
Get Nature+, our best-value online-access subscription
$32.99 / 30 days
cancel any time
Subscribe to this journal
Receive 12 print issues and online access
$259.00 per year
only $21.58 per issue
Buy this article
- Purchase on SpringerLink
- Instant access to full article PDF
Prices may be subject to local taxes which are calculated during checkout
References
Ha, T. et al. Fluorescence resonance energy transfer at the single-molecule level. Nat. Rev. Methods Primers 4, 21 (2024). A review of single-molecule FRET that discusses the benefits and challenges of single-molecule experiments.
Zhou, C. et al. A comprehensive survey on pretrained foundation models: a history from BERT to ChatGPT. Int. J. Mach. Learn. & Cyber. (2024). A review article discussing foundation models, which serve as the template for META-SiM.
Vaswani, A. et al. Attention is all you need. Preprint at arXiv https://doi.org/10.48550/arXiv.1706.03762 (2017). This preprint introduces the attention mechanism that is widely used in large language models and is adapted for single-molecule analysis in META-SiM.
Radford, A. et al. Language models are unsupervised multitask learners. OpenAI blog https://cdn.openai.com/better-language-models/language_models_are_unsupervised_multitask_learners.pdf (2019). This technical report describes the multitask training strategy that underlies modern large language models and that we adapted for single-molecule analysis here.
Semlow, D. R., Blanco, M. R., Walter, N. G. & Staley, J. P. Spliceosomal DEAH-box ATPases remodel pre-mRNA to activate alternative splice sites. Cell 164, 985–998 (2016). This work generated single-molecule FRET datasets monitoring how two molecular machines in yeast spliceosomes act like tiny winches.
Li, J., Zhang, L., Johnson-Buck, A. & Walter, N. G. Automatic classification and segmentation of single-molecule fluorescence time traces with deep learning. Nat. Commun. 11, 5833 (2020). This study introduces AutoSiM as a deep learning tool to achieve rapid, automatic trace selection for improved reproducibility and sensitivity of single-molecule fluorescence microscopy experiments.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This is a summary of: Li, J., Zhang, L., Johnson-Buck, A. & Walter, N. G. Foundation model for efficient biological discovery in single-molecule time traces. Nat. Methods https://doi.org/10.1038/s41592-025-02839-4 (2025).
Rights and permissions
About this article
Cite this article
Machine learning for accelerating discovery from single-molecule data. Nat Methods (2025). https://doi.org/10.1038/s41592-025-02840-x
Published:
DOI: https://doi.org/10.1038/s41592-025-02840-x
