diffdock

1# DiffDock: protein-small-molecule docking
2 
3DiffDock-L generates candidate ligand poses and ranks them by model confidence.
4Confidence is neither a measured probability of correctness nor binding affinity.
5Use this skill for one pair, batches, or separate receptor conformations; treat
6library-wide confidence sorting as pose triage, not hit identification.
7 
8## Verified scope
9 
10Targets the current [v1.1.3 release](https://github.com/gcorso/DiffDock/releases/tag/v1.1.3).
11Released source and current main's identical `inference.py` were checked on
122026-09-30. Bundled helpers were tested on tiny synthetic inputs; pretrained docking,
13ESMFold, CUDA, Docker, GNINA, and hosted-demo execution were **not** run in this
14review. Commands requiring those components are source-verified recipes, not
15successful end-to-end demonstrations.
16 
17## Set up the upstream environment
18 
19```bash
20git clone --branch v1.1.3 --depth 1 https://github.com/gcorso/DiffDock.git
21cd DiffDock
22conda env create --file environment.yml
23conda activate diffdock
24```
25 
26The upstream environment pins Python 3.9.18, CUDA 11.7 Torch/PyG wheels and old
27scientific dependencies. Do not substitute current `torch` or the unrelated PyPI
28`esm` package for `fair-esm`. The published CUDA environment is not a macOS-native
29installation recipe. Follow upstream Docker instructions if suitable:
30 
31```bash
32docker pull rbgcsail/diffdock
33docker run -it --gpus all --entrypoint /bin/bash rbgcsail/diffdock
34micromamba activate diffdock
35```
36…