Evaluating the representational power of pre-trained DNA language models for regulatory genomics
This work highlights a major gap with current gLMs, raising potential issues in conventional pre-training strategies for the non-coding genome.
Authors
- Nikunj V. Somia
- Ziqi Tang (ORCID: https://orcid.org/0000-0001-7585-915X)
- Peter K. Koo (ORCID: https://orcid.org/0000-0001-8722-0038)
- Yiyang Yu (ORCID: https://orcid.org/0009-0001-2925-1909)
Institutions
- Cold Spring Harbor Laboratory (US)
- Columbia University (US)
Publication Details
- Journal
- Genome biology
- Published
- 2025-07-14
- DOI
- https://doi.org/10.1186/s13059-025-03674-8
- Citations
- 17
- Primary Topic
- RNA and protein synthesis mechanisms
- Type
- article
- Field-Weighted Citation Impact
- 13.70
Funders
- National Human Genome Research Institute
- National Institute of General Medical Sciences