Property prediction · Protein Design

ThermoMPNN

Predict the stability change of point mutations in a protein structure.

Runs ThermoMPNN, a ProteinMPNN-derived graph network trained by transfer learning to predict ddG for point mutants. custom_inference.py always scores every substitution at every position of a chain (a saturation scan); a named list of mutations is served by filtering that scan down to the requested substitutions rather than by asking the tool for them specifically, which its own CLI has no option for.

Documentation ↗ Source code ↗

Input

Configure a small job

POST

Job name

Protein PDB

Chain The chain the mutations are scored in; the rest of the structure is context.

Mutations One per line, as wild-type residue, position, and mutant residue, for example T27A. Prefix with a chain and a colon to override the chain above.

Top mutations to highlight Return the K most stabilising substitutions alongside the full table. 0 returns the full table only.

Ready

Response

JSON result

{
  "status": "Ready for a request"
}