Skip to main content

Module sasa

Module sasa 

Source
Expand description

ESM3 SASA (solvent-accessible surface area) tokenization.

Continuous SASA values (Ų) are discretized into 16 bins using the 15 boundary values from esm/utils/constants/esm3.py.

Token layout: 0 = PAD, 1 = MASK, 2 = UNK 3..18 = bins 0..15 (total 19 tokens)

Functions§

tokenize_sasa
Discretize a slice of per-residue SASA values into token IDs.