# gcorso/DiffDock

Implementation of DiffDock: Diffusion Steps, Twists, and Turns for Molecular Docking

Repository: https://github.com/gcorso/DiffDock
Canonical: https://ross.abutalabs.com/products/diffdock
Homepage: https://arxiv.org/abs/2210.01776
Language: Python
License: MIT
License Family: permissive
Topics: diffusion-models, docking, binding, computational-biology, equivariance, machine-learning, non-euclidean-geometry, score-based-models
Last push: 2025-05-02T13:59:19+00:00

## Health v2 (maintenance only)
Score: 31/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 19, release rhythm 8, longevity 100
- inputs: {"age_days": 1429, "days_push": 488, "days_rel": 728, "gap_med": null, "n_releases_24m": 1}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1569, forks 359 (observed 2026-08-28T04:05:05.237733+00:00)

## What it is
DiffDock is a deep learning implementation of a diffusion generative model for molecular docking, predicting how small molecule ligands bind to protein structures. It frames docking as generative modeling over the non-Euclidean manifold of ligand poses (translations, rotations, torsions) and includes the improved DiffDock-L model.

## Use cases
- predict how a small molecule binds to a protein structure
- run molecular docking for drug design
- dock ligands to computationally folded proteins without a known pocket
- generate ligand binding poses with confidence estimates
- replicate state-of-the-art docking benchmarks on PDBBind
- try docking predictions through a web interface

## When to choose
- you need fast, accurate deep-learning-based docking instead of traditional search-based tools
- your protein structure is computationally predicted and lacks a known binding pocket
- you want confidence estimates alongside predicted poses
- you are doing ML research on equivariant diffusion models for molecules

## When to avoid
- you need physics-based scoring or exhaustive conformational search guarantees
- you lack GPU resources or a Python/ML environment
- you require certified binding free-energy calculations rather than pose prediction

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, deep-learning, simulation
- domain: bioinformatics, machine-learning, healthcare, artificial-intelligence
- platform: python
- tags: molecular-docking, diffusion-models, drug-design, equivariance, score-based-models, computational-chemistry, ligand-binding, deep-learning, linux, docker, gpu

## Member repositories
- gcorso/DiffDock (main) score 31

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:05.237733+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T03:58:52.707159+00:00, confidence not recorded.
  - readme: https://github.com/gcorso/DiffDock (fetched 2026-08-28T04:05:05.237733+00:00, sha b218d923b5aa)
  - homepage: https://arxiv.org/abs/2210.01776 (fetched 2026-08-29T11:28:36.564591+00:00, sha 3204511f7004)
  - site_page: https://info.arxiv.org/about/donate.html (fetched 2026-08-29T11:28:36.573949+00:00, sha cca9c3a11c56)
  - site_page: https://info.arxiv.org/about/ourmembers.html (fetched 2026-08-29T11:28:36.577365+00:00, sha 47cbc55ff1de)
  - site_page: https://info.arxiv.org/about (fetched 2026-08-29T11:28:36.579214+00:00, sha a1f16f915a9a)
  - site_page: https://info.arxiv.org/labs/index.html (fetched 2026-08-29T11:28:36.575698+00:00, sha b14a8d05a0ec)
- Data as of 2026-08-30T08:39:29.467469+00:00.