Predict Structure Boltz2
predict_structure_boltz2Read-only
Predict biomolecular structures: use direct multimolecular prediction for protein, ligand, DNA, RNA complexes, or a UniProt-to-docking workflow for protein-ligand complexes.
Instructions
Boltz-2 structure workflow. Use mode='structure' for direct multimolecular structure prediction or mode='protein_ligand' for the integrated UniProt-to-docking workflow.
Input Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| mode | No | Boltz-2 workflow mode. | structure |
| protein_sequences | No | Protein sequences for direct Boltz-2 prediction. | |
| ligand_smiles | No | Ligand SMILES strings. | |
| dna_sequences | No | Optional DNA sequences for complex prediction. | |
| rna_sequences | No | Optional RNA sequences for complex prediction. | |
| uniprot_accession | No | UniProt accession used when mode='protein_ligand'. | |
| predict_affinity | No | Predict ligand binding affinity when ligands are present. | |
| method_conditioning | No | Optional structure-style conditioning. | x-ray |
| pocket_residues | No | Optional binding-pocket residue constraints. | |
| recycling_steps | No | Boltz-2 recycling iterations. Default 3. | |
| sampling_steps | No | Diffusion sampling steps. Default 200. | |
| diffusion_samples | No | Number of structural hypotheses. Default 1. |