drugforge

A toolkit for structure-based open antiviral drug discovery based on the work by the ASAP Discovery Consortium.

This project is a fork of ASAPDiscovery by the ASAP Discovery Consortium. We thank the original authors for their work and for making it available under MIT license.

This fork is now developed and maintained independently by the Chodera Lab.

Intro

All pandemics are global health threats. Our best defense is a healthy global antiviral discovery community with a robust pipeline of open discovery tools.

The toolkit in this repo is a batteries-included drug discovery pipeline being actively developed in a transparent open-source way, with a focus on computational chemistry and informatics support for medicinal chemistry.

Getting Started

Install the drugforge subpackages and begin to explore! The drugforge documentation can be found here.

There are a range of workflows and tooling to use split into several namespace subpackages by theme.

Warning: The implementation of drugforge-ML is still in the works and will be available in the next release. If you want to use ML scoring please refer to the original version of this code under asapdiscovery.

drugforge-alchemy: Free energy calculations using OpenFE and Alchemiscale. See tutorial and CLI guide

drugforge-cli: Command line tools uniting the whole repo.

drugforge-data: Core data models and integrations with services such as Postera.ai. See tutorial

drugforge-dataviz: Data and structure visualization using 3DMol and PyMOL. See tutorial

drugforge-docking: Docking and compound screening with the OpenEye toolkit

drugforge-spectrum: Working with sequence and fitness information. See tutorial

drugforge-ml: Structure based ML models for predicting compound activity. See tutorial and CLI guide

drugforge-modelling: Structure prep and standardisation

drugforge-simulation: MD simulations and analysis using OpenMM. See tutorial

drugforge-workflows: Workflows that combine components to enable project support. See tutorial and CLI guide

Disclaimer

drugforge is pre-alpha and is under very active development, we make no guarantees around correctness and the API is liable to change rapidly at any time.

Installation

Note: currently all drugforge packages support Python <=3.11 only.

Installing from conda-forge (WIP)

You can install the package from the wonderful conda-forge with mamba install -c conda-forge drugforge.

Developer install

drugforge is a namespace package, composed of individual Python packages with their own dependencies. Each of these packages follows the drugforge-* convention for the package name, e.g. drugforge-data.

To install an drugforge package hosted in this repository, we recommend the following:

1. Clone the repository, then enter the source tree:

   git clone https://github.com/choderalab/drugforge.git cd drugforge

2. Install the dependencies into a new conda environment, and activate it: NOTES: Conda will almost certainly fail to build the environment - mamba is a drop-in replacement for conda that is much faster and more reliable. Additionally, if the environment is built on a CPU, torch may not compile with GPU support. Instead, build the environment as described on a GPU node; the architecture will be detected automatically. Alternatively to build for a specific CUDA version you can use the

   mamba env create -f devtools/conda-envs/drugforge-{platform}.yml conda activate drugforge Alternatively to build for a specific CUDA version you can use the following. export CONDA_OVERRIDE_CUDA=12.2 && mamba env create -f devtools/conda-envs/drugforge-{platform}.yml && conda activate drugforge

3. Install the individual drugforge packages you want to use with pip into the environment. For example, drugforge-data:

   pip install drugforge-data

Contributing

pre-commit

We use pre-commit to automate code formatting and other fixes. You do not need to install pre-commit as we run it on our CI. If you want to run it locally: ```bash

install

$ mamba install -c conda-forge pre-commit

check

$ pre-commit --version pre-commit 3.0.4 # your version may be different $ pre-commit install ```

Now every time you make a commit, the hooks will run on just the files you changed. See here for more details.

Copyright

Copyright (c) 2023, ASAP Discovery Copyright (c) 2025, Chodera Lab

Acknowledgements

Project based on the Computational Molecular Science Python Cookiecutter version 1.6.