Preprocessing file generator for Protein 1 + Ligand + Protein 2 binding prediction with Boltz-2. Drop in a Protein 1 + Ligand docked structure and a Protein 1 + Protein 2 reference structure — BoltzYML extracts sequences, identifies the ligand and pocket residues, and emits a ready-to-run YAML for the Boltz-2 CLI.

BoltzYML pipeline Two CIF inputs flow through parsing, ligand and chain detection, pocket contact computation, and YAML emission, producing a Boltz-2 input file. INPUTS Protein 1 + Ligand docked reference .cif Protein 1 + Protein 2 reference structure .cif PIPELINE (in browser) 1 Parse CIFs atoms · chains · residues 2 Detect ligand + extract sequences 3 Pocket contacts CA atoms ≤ cutoff Å 4 Emit YAML v1 chains A · B · C OUTPUT job.yaml Then run locally: boltz predict job.yaml --use_msa_server --use_potentials --diffusion_samples 3

1 · Ligand-bound CIF (Protein A + Ligand)

2 · Complex CIF (Protein A + Protein B)

Options

The detected ligand code from the CIF goes into the YAML's ccd: field. If your CIF labels the ligand generically (e.g. LIG, UNL, UNK), type the real PDB CCD code here to replace it — e.g. A8S for abscisic acid, ATP, HEM. Whatever you type is used verbatim.

Detection summary

Status
Waiting for both CIF files…
How chain assignment works

Protein A (Boltz chain A) is the protein in the ligand CIF. Protein B (Boltz chain C) is matched in the complex CIF by sequence similarity to Protein A — the remaining chain becomes Protein B. Residue numbers use auth_seq_id from the ligand CIF.

Output YAML