Foundation models of protein sequences: A brief overview

Andreas Bjerregaard, Peter Mørch Groth, Søren Hauberg, Anders Krogh, Wouter Boomsma

Abstract: Highlights•A review of three decades of statistical protein modeling.•The key difference between models lies in how they model amino acid context.•Protein foundation models are becoming increasingly multi-modal.•Fine-tuned models outperform fixed protein representations.•Further scaling shows tendencies of diminishing returns.

External IDs:doi:10.1016/j.sbi.2025.103004