data/README.md

Example data

Training dataset

CSV file with columns mutation and property_value. Variants should be formatted as A40P/E61Y, or for protein complexes as A40P/E61Y:WT, where : separates the individual chains (e.g. chain 1 mutations:chain 2 mutations), / separates the individual mutations, and WT indicates the wildtype sequence.

Multi-chain protein

The variable heavy domain (vh) and variable light domain (vl) chains of HuABC2, anti-CD122 antibody.

Single chain protein

Engineered ascorbate peroxidase (APEX)