Tools

Plex is paving the way for permissionless science by ensuring that computational biology tools are not just available, but also easily accessible for open-source, early-stage drug discovery.

To facilitate ease of access and the spirit of open science, plex uses Docker containers of computational biology tools and makes them publicly available. These tools enable transparency and easy replication, ensuring that researchers can validate, reproduce, and build upon existing work with confidence.

Tool Configs

Plex employs tool configs as computation templates which dictate how computations should be carried out. As demonstrated in the JSON below, these configs

• Specify the Docker container used
• Detail the input data format, ensuring that the data fed into the tool aligns with its expectations
• Define the output data format, allowing for standardized retrieval and further processing

This approach, reminiscent of the Common Workflow Language (CWL), ensures consistency, interoperability, and reproducibility across different tools and workflows.

Colabfold Tool Config

{
    "class": "CommandLineTool",
    "name": "colabfold-mini",
    "description": "Protein folding prediction using Colabfold (mini settings)",
    "baseCommand": ["/bin/bash", "-c"],
    "arguments": [
      "colabfold_batch --templates --max-msa 32:64 --num-recycle $(inputs.recycle.default) /inputs /outputs;"
    ],
    "dockerPull": "public.ecr.aws/p7l9w5o7/colabfold:latest",
    "gpuBool": true,
    "networkBool": true,
    "inputs": {
      "sequence": {
        "type": "File",
        "item": "",
        "glob": ["*.fasta"]
      },
      "recycle": {
        "type": "int",
        "item": "",
        "default": "1"
      }
    },
    "outputs": {
      "best_folded_protein": {
        "type": "File",
        "item": "",
        "glob": ["*rank_1*.pdb"]
      },
      "all_folded_proteins": {
        "type": "Array",
        "item": "File",
        "glob": ["*rank*.pdb"]
      }
    }
}