Rosettafold3 Tools#

Configuration File: remote_tools/rosettafold3_tools.json Tool Type: Remote Tools Count: 1

This page contains all tools defined in the rosettafold3_tools.json configuration file.

Available Tools#

rosettafold3_predict (Type: RemoteTool)#

Predict the all-atom structure of a biomolecular assembly with RoseTTAFold3 and return the per-sa…

rosettafold3_predict tool specification

Tool Information:

  • Name: rosettafold3_predict

  • Type: RemoteTool

  • Description: Predict the all-atom structure of a biomolecular assembly with RoseTTAFold3 and return the per-sample confidence scores (pTM, ipTM, ranking score) with the paths to the predicted mmCIF models. Input is RoseTTAFold3’s own specification as text: a JSON list of items, each {“name”: …, “components”: [{“seq”: …, “chain_id”: …, “chain_type”: …}]}, or a CIF/PDB structure to re-predict. No MSA server is used; set LOCAL_MSA_DIRS on the worker or give a component an msa_path to fold with MSAs.

Parameters:

  • spec (string) (required) RoseTTAFold3 input specification. Example:

[{“name”: “barstar”, “components”: [

{“seq”: “KKAVINGEQIRSISDLHQTL…”, “chain_id”: “A”, “chain_type”: “polypeptide(L)”}]}]

chain_type is a PDB polymer type: ‘polypeptide(L)’, ‘polyribonucleotide’, ‘polydeoxyribonucleotide’.

  • seed (integer) (optional) Random seed; the same seed reproduces the fold.

  • annotate_b_factor_with_plddt (boolean) (optional) Write per-atom pLDDT into the mmCIF B-factor column (default false).

  • dump_trajectories (boolean) (optional) Also write the diffusion trajectory of each sample (default false; large).

  • one_model_per_file (boolean) (optional) Write each sampled model to its own file (default false).

  • add_missing_atoms (boolean) (optional) When the input is a CIF/PDB structure, add atoms missing from its residues using the CCD (default true; no effect on a JSON specification).

  • return_structures (boolean) (optional) Return the mmCIF text of every sampled structure (default false: five models of a real assembly are megabytes, and a reply that large is what a transport truncates into invalid JSON). The scores come back either way; the paths do too, but they name a temporary directory this call removes on return, so set this to keep the coordinates.

Example Usage:

query = {
    "name": "rosettafold3_predict",
    "arguments": {
        "spec": "example_value"
    }
}
result = tu.run(query)