Mosaic Tools#

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

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

Available Tools#

mosaic_design_binder (Type: RemoteTool)#

Design a protein binder against a target protein by gradient descent through Boltz-2 (escalante-b…

mosaic_design_binder tool specification

Tool Information:

  • Name: mosaic_design_binder

  • Type: RemoteTool

  • Description: Design a protein binder against a target protein by gradient descent through Boltz-2 (escalante-bio/mosaic). Give the target’s amino-acid sequence and a binder length; it invents the binder: 125 optimiser steps, each a full Boltz-2 forward+backward pass under a contact + inverse-folding loss, then one refold to score the result. Returns the binder sequence, ipTM, pLDDT and binder-to-target PAE, and (opt-in) the designed complex as PDB text. Single-sequence: no MSA is built or fetched. A request takes minutes; the first request of a new size also pays a compilation of minutes.

Parameters:

  • target_sequence (string) (required) The target protein, one chain, standard amino acids only.

  • binder_length (integer) (optional) Residues in the binder to invent.

  • target_copies (integer) (optional) Copies of the target chain in the complex (for a homo-oligomeric target).

  • epitope (string) (optional) Optional target residues the binder must contact, 1-based along target_sequence, e.g. ‘12,15,40-48’; applied on every copy.

  • seed (integer) (optional) Random seed for the starting sequence and the optimiser. Vary it for different designs.

  • steps1 (integer) (optional) Optimiser steps in stage 1 (upstream: 75). Time is linear in the total of steps1 + steps2.

  • steps2 (integer) (optional) Optimiser steps in stage 2 (upstream: 50).

  • return_structure (boolean) (optional) Return the designed complex as PDB text (default false).

Example Usage:

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