Docking & Binding · MIT / Barzilay Lab
Diffusion docking: 38% top-1 success versus 23% for classical AutoDock.
30s–5min
Runs on Managed GPU endpoint
Runs on managed compute. Availability and startup time depend on current capacity.Run this model
About this model
DiffDock treats docking as generative modelling over ligand poses rather than search-and-score. It nearly doubles top-1 accuracy over AutoDock Vina and, critically, holds up on computationally predicted structures — which is what you usually have.
Reported performance
Standardized I/O contract
Every model in the Hub speaks the same contract, which is what lets the Router and the agent call any of them without special-casing.
Inputs
Receptor (PDB)
Ligand SMILES
Poses
Outputs
All-atom coordinates in PDB format, viewable and downloadable.
Poses with confidence score and RMSD spread.
