boltz
boltz copied to clipboard
Bond constraints not working
We ran the boltz prediction using part of the yaml of the docstring in the yaml.py file:
sequences:
- protein:
id: A
sequence: "MADQLTEEQIAEFKEAFSLF"
- protein:
id: [B, C]
sequence: "AKLSILPWGHC"
constraints:
- bond:
atom1: [A, 1, CA]
atom2: [A, 2, N]
The bond constraint did not change the distance between the atoms. Furthermore, the bond constraints are not listed in the file boltz_results_test2/processed/manifest.json as shown below:
{"records": [{"id": "test2", "structure": {"resolution": null, "method": null, "deposited": null, "released": null, "revised": null, "num_chains": 3, "num_interfaces": null}, "chains": [{"chain_id": 0, "chain_name": "A", "mol_type": 0, "cluster_id": -1, "msa_id": "test2_0", "num_residues": 20, "valid": true, "entity_id": 0}, {"chain_id": 1, "chain_name": "B", "mol_type": 0, "cluster_id": -1, "msa_id": "test2_1", "num_residues": 11, "valid": true, "entity_id": 1}, {"chain_id": 2, "chain_name": "C", "mol_type": 0, "cluster_id": -1, "msa_id": "test2_1", "num_residues": 11, "valid": true, "entity_id": 1}], "interfaces": [], "inference_options": {"binders": [], "pocket": []}}]}
If we measure the distance between the 2 constrained atoms in the predicted structure, we get a distance of 2.4 A instead of about 1.5 A.
We also tried to define disulfide bridge in the bond constraint section, without any success.
Thanks for your help, Best,