Larger PLMs generate more novel sequences from more sparsely populated protein families

Summary

Larger protein language models generate more novel protein sequences from sparsely populated families (1). These sequences are also slightly more likely to express, although the effect isn’t huge. These results were observed using ProGen3.

Figures

Ref (1)

See also

• Sequence-only protein language models implicitly cluster proteins at fineness levels that increase with size

1.

Bhatnagar A, Jain S, Beazer J, Curran SC, Hoffnagle AM, Ching KS, et al. Scaling Unlocks Broader Generation and Deeper Functional Understanding of Proteins. openRxiv; 2025. Available from: https://doi.org/10.1101/2025.04.15.649055