Esm If1 Tools#

Configuration File: remote_tools/esm_if1_tools.json Tool Type: Remote Tools Count: 1

This page contains all tools defined in the esm_if1_tools.json configuration file.

Available Tools#

Design protein SEQUENCES for a given backbone with ESM-IF1 (inverse folding). Give the backbone a…

esm_if1_design tool specification

Tool Information:

Name: esm_if1_design
Type: RemoteTool
Description: Design protein SEQUENCES for a given backbone with ESM-IF1 (inverse folding). Give the backbone as PDB or mmCIF text and, optionally, the chain to design; returns num_samples sampled sequences, the structure’s own sequence, and each sample’s sequence recovery against it. Raise ‘temperature’ for diversity. Use multichain_backbone=true to design one chain of a complex while conditioning on all of its backbones. Runs on the CPU: expect ~1.5 s for a 64-residue backbone and ~40 s for 800 residues, roughly linear in length.

Parameters:

structure (string) (required) The backbone, as the text of a PDB or mmCIF file. Only the N, CA and C atoms of amino-acid residues are read; side chains, waters and ligands are ignored. Up to 8 MB.
structure_format (string) (optional) Which of the two the ‘structure’ text is (default pdb). Upstream’s loader dispatches on this, so a cif sent as pdb fails to parse rather than being sniffed.
chain (string) (optional) The chain id to design. Omit to take every chain the file’s loader returns as one backbone. Required when multichain_backbone is true.
num_samples (integer) (optional) How many sequences to sample (default 1, max 64). Samples are independent draws from the same conditional distribution.
temperature (number) (optional) Sampling temperature (default 1.0). Higher is more diverse and lower recovers the native sequence more often.
seed (integer) (optional) Random seed, applied once per request (default 37, the kit’s own). At a fixed seed, num_samples and batch_size the sampled sequences repeat exactly.
batch_size (integer) (optional) How many samples share one forward pass (default 64, max 64). It changes speed and memory, not the distribution. min(batch_size, num_samples) x the backbone’s length must not exceed 6400.
multichain_backbone (boolean) (optional) Design ‘chain’ conditioned on the backbones of every chain in the file (upstream’s multichain route) rather than on that chain alone. Needs ‘chain’.

Example Usage:

query = {
    "name": "esm_if1_design",
    "arguments": {
        "structure": "example_value"
    }
}
result = tu.run(query)