Boltz-2

New
Structure Prediction ~10mvBoltz 2.2.1

AlphaFold3 reproduction, predict protein/nucleic acid/small molecule complexes

Example Results API Docs
Examples
8JYR - Antibody complex
7QF4 - Protein complex
Protein-Ligand Complex + Affinity
Copies

1

Chain A

Cyclic

Use YAML file input

Upload a Boltz YAML configuration file instead of filling out the UI. MSA entries are not required — MSAs are computed and injected automatically for protein chains.


Number of Recycles

Number of rounds of refinement during structure prediction. Higher values may improve accuracy but take longer.


Step Scale

Lower step scale means more diversity (recommended between 1 and 2).


Number of Seeds

Number of jobs to submit. Each job uses a random seed by default.


Seed (optional)

Set a specific seed for reproducibility, or leave blank for a random seed.


Output Format

File format used for predicted structures.


Boltz Version

Which Boltz model version to use for prediction.


Method

Experimental method bias for the prediction (optional).


Template CIFs

Upload template CIF files to guide the structure prediction. These will be used as templates for the prediction.


Template Distance Threshold

Controls the distance (in Angstroms) that the prediction can deviate from the template.


Use Potentials

Run the original Boltz-2 model using inference time potentials. Significantly improves the physical quality of the poses.


Use Multiple Sequence Alignment

When disabled, Boltz will not use MSA data for prediction.


Custom A3M Files

Upload custom a3m (multiple sequence alignment) files to use instead of generating MSAs automatically. Each a3m file should correspond to a protein chain in your complex. MSA generation is skipped for chains you provide an a3m for.


Max MSA Sequences

Max number of sequences to use for MSA. Use a smaller value for a more shallow MSA, which may produce predictions closer to your template.

MSA Database

Sequence database to search when building the MSA. UniRef30 searches are extended with the ColabFold environmental database (BFD/MGnify-derived). SwissProt restricts the search to curated SwissProt sequences.


Generate IPSAE metrics

Generate IPSAE metrics for protein-protein complex interfaces.

All sequences must be non empty.





Structure Prediction