Deep-learning prediction of gene expression from personal genomes
Our work suggests that fine-tuning with personal genomes corrects previously reported shortcomings of gene expression prediction across unseen individuals, but does not learn a regulatory grammar that generalizes to unseen loci. Fine-tuned deep expression models thus share similar performance and limitations of state-of-the-art linear models, highlighting a gap for the field.
Authors
- Katherine S. Pollard (ORCID: https://orcid.org/0000-0002-9870-6196)
- Sean Whalen (ORCID: https://orcid.org/0000-0002-6648-3610)
- Shiron Drusinsky (ORCID: https://orcid.org/0000-0003-1184-8835)
Institutions
- Gladstone Institutes (US)
- University of California, San Francisco (US)
- Chan Zuckerberg Initiative (United States) (US)
- University of San Francisco (US)
Publication Details
- Journal
- Genome biology
- Published
- 2026-01-06
- DOI
- https://doi.org/10.1186/s13059-025-03926-7
- Primary Topic
- Genomics and Chromatin Dynamics
- Type
- article
- Field-Weighted Citation Impact
- 0.00