LigandMPNN
Design protein sequences while accounting for ligands (binding partners) and modified residues.
API/CLI tool ID: ligandmpnn
Understanding inputs and outputs
LigandMPNN proposes amino acid sequences using a protein structure and surrounding ligands. A ligand is a molecule that can bind to a protein. The design uses ligand positions in the input structure; it does not generate new ligands or search for binding sites.
designedResidues selects the amino acid positions that may change, grouped by chain. Use PDB residue numbers from the input structure, including insertion codes where present, rather than positions counted along the sequence. Unselected positions retain their original amino acids. An empty object makes all positions eligible for design.
overall_confidence scores the sequence across all designed positions, while ligand_confidence scores the subset evaluated near ligands. Use them to compare candidates under the same input and evaluation conditions, not as binding strengths or probabilities of experimental success. When ligand_residues is 0, no positions are available for the local score and ligand_confidence is empty. This differs from a candidate receiving a low score.
Input settings
| Setting | Format | Required | Default | Description and constraints |
|---|---|---|---|---|
pdbFile | pdb | Required | — | Extensions pdb · cif |
designedResidues | selectMultichainResidues | Required | {} | PDB residue numbers per chain. Only the listed positions are designed and the rest stay fixed. An empty object designs everything. |
numSequences | number | Optional | 8 | Minimum 1, Maximum 100 |
Output metrics
| Column | Display name | Range | Description |
|---|---|---|---|
overall_confidence | Designed sequence score | — | Model score: exp(−x) of the mean negative log probability over designed positions. Higher ranks first. Not binding affinity or the probability of experimental success. |
ligand_confidence | Design score near the ligand | — | Model score for evaluated positions near the ligand. Blank if no residues were evaluated. |
ligand_residues | Residues evaluated near the ligand | — | Number of residues used for the score near the ligand. 0 means the structure has no ligand or no nearby residues, so that score is blank. |
sequence | Designed sequence | — | Protein sequence proposed by the model (one-letter amino acids). A computed proposal, not a sequence confirmed to express or function. |
See Reading results for metric definitions and interpretation. A range shows possible values; whether higher or lower is better depends on the metric.
Execution availability
Open an execution plan in your project to see available tool versions and costs. Review and approve the inputs, settings, and terms before running.
License
Code, model weights, and accompanying data have their own terms. The table summarizes terms from official sources; check your execution plan for the configuration you will use. License texts are available in Third-party notices.
| Component | License | Commercial use | Source | Notes |
|---|---|---|---|---|
| Code | MIT | Allowed, subject to notice requirements | License text | — |
| Model weights | MIT | Allowed, subject to notice requirements | Repository README | The README states that both code and model parameters are MIT licensed |
Checked on 2026-10-06.