ProteinMPNN designs sequences for a fixed backbone by maximizing sequence-backbone compatibility via message passing. Widely used for inverse folding and binder design.
Type
Inverse folding (sequence design)
Input
Backbone coordinates (PDB)
Output
Sequences consistent with backbone
Origin
Baker Lab
Resources
- GitHub: dauparas/ProteinMPNN
- Paper: Robust deep learning–based protein sequence design using ProteinMPNN
Pair with structure predictors (e.g., ESMFold/AlphaFold2) to validate designed sequences.