Boltz-2
Cofold validation. Validate designed binders against your antigen with an antibody-trained cofold model. ~15 s per design in single-sequence mode.
What it is for
Pick Boltz-2 to validate a designed binder against your antigen. Single-sequence cofold with interface confidence (ipTM), antibody-trained and orthogonal to AF2-multimer. For sequence design, use ProteinMPNN; for de novo backbones, use RFantibody, BindCraft, or BoltzGen first.
Boltz-2 (Wohlwend et al., bioRxiv 2025). An open-weights structure prediction model trained on antibody-antigen complexes with a calibrated confidence head. Single-sequence mode is orthogonal to AF2-multimer: when both agree, the predicted complex is real; when they disagree, the disagreement itself is informative. Returns a folded complex PDB plus ipTM, pTM, and complex_pLDDT per design.
When it fits:
- You designed binders with MPNN, RFantibody, BindCraft, BoltzGen, RFdiffusion, or PXDesign and need to score them against the intended antigen.
- You have native or near-native scFv / Fab / nanobody / peptide sequences you want a fast independent fold for.
- AF2-multimer ipTM is saturated and you want a second confidence channel from a different architecture before ordering DNA.
Inputs
You will need:
- Antigen PDB or mmCIF (single chain; the binder is added separately).
- One or more binder sequences (scFv, nanobody, peptide, anything that folds as a single protein chain), 20 to 400 aa each.
- Optional: a list of antigen hotspot residue numbers to count contacts against (1-indexed on the chosen antigen chain).
Each run uses a preset that sets the scale and scope:
- Single-sequence (fast)
- YAML ``msa: empty`` per chain. The right choice for designed sequences (MPNN, RFantibody, BindCraft, BoltzGen, RFdiffusion, PXDesign outputs) where no informative MSA exists. ~15 s/design on A100-40GB.
- With MSA (slower, natural sequences)
- Boltz fetches MSAs from the public ColabFold MMseqs2 endpoint at runtime. Better for natural / near-native sequences; ~3 min/design including MSA fetch.
Parameters you set on the form:
- Antigen PDB
- Upload the target structure as .pdb, .cif, or .mmcif. CIF inputs are converted to PDB server-side.
- Antigen chain
- Single chain ID (e.g.
A) that Boltz-2 should treat as the antigen. The binder folds as a separate chain. - Hotspot residues
- Optional. Comma-separated 1-indexed positions on the antigen chain (e.g.
55,56,57,71,72,73,74). The pipeline reports how many of these residues the binder contacts (heavy atom within 5 Å). - Binder sequences
- Paste one sequence per line, or upload as FASTA (
>nameheaders). Each sequence folds independently against the antigen. 20 to 400 aa per binder, up to 50 binders per run. - Preset
- Single-sequence (default) folds in
msa: emptymode, the right choice for designed sequences. With MSA fetches MSAs from the public ColabFold MMseqs2 endpoint and is slower but more accurate on natural sequences.
Typical runtime:
- standalone
- <1 min/design
- msa_server
- ~3 min/design
How to read the results
Per-design folded complex PDB + ipTM, pTM, complex_pLDDT, complex_iplddt, and hotspot contact count. Strict-pass classification (complex_pLDDT > 0.85, ipTM > 0.7, n_hotspot_contacts > 4) surfaces which designs are worth ordering.
Where a tool reports them, the scores mean:
- ipTM
- Predicted confidence in the binder to target interface. Higher is better. Aim above roughly 0.7 on a tractable target.
- pLDDT
- Per-residue confidence in the predicted fold. Higher means the model is more sure of that part of the structure.
- i_pAE and pAE
- Predicted alignment error, at the interface (i_pAE) or across the whole structure (pAE). Lower is better.
References
Wohlwend et al., bioRxiv 2025