ProteinMPNN & LigandMPNN
Design protein or antibody sequences for a structured complex.
Results
Loading starting example…
Method & evidence
Choose ProteinMPNN, SolubleMPNN, LigandMPNN, or the antibody-specific AbMPNN model; generate sequence candidates for a full backbone or atomic-context complex, preserve chosen positions, and compare recovery, mutations, and model confidence.
Scope & limitations
- Standard ProteinMPNN includes vanilla and soluble full-backbone models at four training-noise levels.
- LigandMPNN requires non-protein atomic context and uses its own models and input rules.
- AbMPNN is a distinct antibody-specific model, not a standard ProteinMPNN checkpoint.
- CA-only, PSSM, tied-position, residue-bias, and probability-matrix modes are not available in this analysis.
Interpreting results
- ProteinMPNN score is a negative-log-probability measure; lower values indicate sequences the model considers more compatible with the backbone.
- LigandMPNN overall and ligand-context confidence values are model probabilities for ranking candidates, not measurements of binding affinity or catalytic activity.
- Sequence recovery is similarity to the source sequence, not an independent quality score. Interpret it in light of the intended redesign depth.
- AbMPNN is an antibody-specific prediction model, not an experimental binding or developability assay. Compare candidates only after confirming the heavy/light roles and fixed-position constraints.
- Review chain-resolved mutations and fixed-position compliance before advancing a candidate to structure prediction or experimental work.
References
Input requirements & outputs
Input checks
- Complete full-backbone PDB coordinates
- Standard method: one or more designable protein chains
- LigandMPNN: non-water atomic context for a ligand, nucleic acid, metal, cofactor, or other molecule
- AbMPNN: one heavy chain and an optional distinct light chain
- Up to 2 structures and 2,000 total residues per analysis
Expected results
- Ranked sequence designs — Compare model score or confidence, atomic-context confidence where applicable, and sequence recovery.
- Chain-resolved sequences and roles — Review designed chains without losing fixed context; AbMPNN results retain explicit heavy/light roles.
- Exact mutation list — Inspect residue changes using both sequence and PDB residue positions.
