diffdock

What problem does it solve?

DiffDock predicts ligand-binding poses to protein targets using diffusion-based models, generating multiple candidate poses with confidence scores for structure-based drug design and virtual screening.

Core Features & Use Cases

• Pose prediction for protein–ligand complexes
• Confidence scores per pose to prioritize candidates
• Batch docking and ESEmbeddings-based acceleration
• Supports PDB structures or protein sequences via ESMFold

Quick Start

Example: run single-pose docking for a protein and SMILES ligand; output ranks and confidences.