BioinformaticsbioRxivPreprint
✦ In-depth digest
SaProt: Protein Language Modeling with Structure-aware Vocabulary
Jin Su, Chenchen Han, Yuyang Zhou…·Fajie Yuan
Pairing each residue with its Foldseek 3D-structure token and pre-training an ESM-2-sized model on about 40 million AlphaFold2 structures beat ESM-2 and structure-based baselines on variant-effect and function tasks.
244 citations