DiffDock
BMIT
Molecular Dockingopen
Updated 3 months agoNextIn take
DiffDock was the first to show that diffusion models could do molecular docking — a genuinely important contribution — but the field has moved past it. Boltz-2 does the same thing better and in a unified model.
Specifications
| Architecture | Diffusion generative model over ligand poses in the SE(3) space |
| Parameters | ~50M |
| Training Data | PDBBind refined set, PDB protein-ligand complexes |
| License | MIT license. Fully open weights and code. |
| Hardware | 1x GPU with 16GB+ VRAM |
| Inference Cost | Self-hosted — ~$0.005/docking on cloud GPU |
| API Available | No |
| Weights Available | Yes |
Benchmark Performance
| Benchmark | Score |
|---|---|
| PoseBusters | 58% valid |
| PDBBind | 38% top-1 < 2Å |
When to Use This
- +Blind docking where the binding site is unknown
- +Research pipelines needing a fully open docking method
When NOT to Use This
- −Production virtual screening where accuracy is paramount — Boltz-2 is better
- −When you know the binding site — classical methods may suffice
Production Readiness
research
Known Users
- Academic drug discovery labs
Grade Rationale
B
Pioneered the diffusion approach to molecular docking and remains a solid baseline. But Boltz-2 and AF3 have since surpassed it on accuracy, and the PoseBusters pass rate shows room for improvement on physical validity.
Sources
Update History
2026-01-10Initial entry