SUBSEQ.BIO
DOCS-BOLTZGEN

BoltzGen

Generate candidate protein or peptide binders for protein, peptide, or small-molecule targets.

Overview

  • Early-stage binder candidate generation for a known protein or peptide target.
  • Short peptide or cyclic peptide binder exploration.
  • Protein binder design around a small molecule described by CCD code or SMILES.
  • Creating a ranked candidate set for manual inspection and downstream validation.

Modes

ModeInput shapeWhen to use it
protein_binder
Protein Binder Default		 Uses one selected file/source when file parameters are present. Design a protein binder against a protein or peptide target structure.
peptide_binder
Peptide Binder		 Uses one selected file/source when file parameters are present. Design a linear or cyclic peptide against a protein target structure.
small_molecule_binder
Small-Molecule Binder		 No uploaded input is required by the mode itself. Design a protein binder against a CCD or SMILES small molecule.

Canonical Job Configuration

These are the fields exposed by the default job configuration for boltzgen. They are also returned by GET /api/v1/program/params?program=boltzgen and submitted as the params JSON object to POST /api/v1/job/submit.

ParameterTypeModesWhat it does
target_structure
Target Structure		 Structure file Protein Binder, Peptide Binder Choose a PDB or mmCIF target from an upload, dataset, or previous job output.
Required; Files: .pdb, .cif, .mmcif
target_chains
Target Chains		 Text Protein Binder, Peptide Binder Use all chains, or provide the target chain IDs to include.
Default: all
binding_site
Binding Site		 Residue selection Protein Binder, Peptide Binder Optional target residues to prefer for binding. Use chain:residues unless exactly one target chain is selected.
protein_binder_length
Binder Length		 Length or range Protein Binder Protein binder length or range.
Default: 80-140
peptide_length
Peptide Length		 Length or range Peptide Binder Peptide length or range.
Default: 12-20
cyclic_peptide
Cyclic Peptide		 Yes/no Peptide Binder Request a cyclic peptide design.
Default: false
small_molecule_binder_length
Binder Length		 Length or range Small-Molecule Binder Protein binder length or range for the small-molecule target.
Default: 140-180
ligand_type
Ligand Type		 Text Small-Molecule Binder Describe the small molecule by Chemical Component Dictionary code or SMILES.
Default: CCD Code; Options: CCD Code, SMILES
ligand_ccd
CCD Code		 Text Small-Molecule Binder TSA
Required; Shown when ligand_type is CCD Code
ligand_smiles
SMILES		 Ligand text Small-Molecule Binder CC(=O)NCCNC(C)=O
Required; Shown when ligand_type is SMILES
num_designs
Generated Designs		 Integer All modes Number of intermediate designs to generate before filtering.
Default: 50; Range: 1-60000
final_designs
Final Designs		 Integer All modes How many ranked, diversity-selected designs to keep.
Default: 10; Range: 1-1000
Advanced configuration fields
ParameterTypeModesWhat it does
inverse_fold_sequences
Sequences Per Backbone		 Integer All modes
Default: 1; Range: 1-32
selection_focus
Selection Focus		 Text All modes Tune final selection toward model quality or sequence diversity.
Default: Protocol Default; Options: Protocol Default, More Quality, More Diversity

Outputs And Metrics

  • Ranked final structure files.
  • Metrics tables for considered and selected designs.
  • An overview PDF with ranking and metric plots when available.
  • Final rank is a composite ranking after filters and diversity selection.
  • Refolding RMSD compares designed structure after re-prediction; lower is better.
  • Interface contacts, buried area, and confidence/interface metrics are screening signals, not proof of binding.

Common Examples

  • Protein binder: target chain A, binding site A:45-53,72, binder length 80-140, keep 10 final designs.
  • Linear peptide binder: peptide length 12-20 with a focused epitope.
  • Small-molecule binder: CCD code or SMILES, binder length 140-180, small final set for inspection.

Example API params

{
  "mode": "protein_binder",
  "target_structure": "target.pdb",
  "target_chains": "A",
  "binding_site": "A:45-53,72",
  "protein_binder_length": "80-140",
  "num_designs": 50,
  "final_designs": 10
}

Caveats

  • Outputs are computational candidates, not validated binders.
  • Verify chain IDs and residue numbering before defining a binding site.
  • For SMILES, encode intended stereochemistry, protonation, and tautomer state when relevant.

Advanced Submit

Advanced submit is still available for direct program arguments through POST /api/v1/job/submit-advanced. Prefer canonical configuration unless you need exact low-level arguments or are reproducing a known command line.

  • Advanced submit is useful for reproducing a lower-level BoltzGen config or accessing fields not exposed in canonical configuration.
  • Use canonical configuration for routine protein, peptide, and small-molecule binder runs.
curl -X POST https://subseq.bio/api/v1/job/submit \
  -H "Authorization: Bearer <api_key>" \
  -F program=boltzgen \
  -F 'params={"mode":"protein_binder","target_structure":"target.pdb","target_chains":"A","binding_site":"A:45-53,72","binder_length":"80-140","generated_designs":50,"final_designs":10}'

Further Reading