alterlab-diffdock

What problem does it solve?

This DiffDock-based skill provides a diffusion-model workflow to predict 3D protein-ligand binding poses and associated confidence scores, enabling structure-based interpretation and downstream rescoring.

Core Features & Use Cases

• Pose prediction: generate candidate binding poses for protein targets and ligands (SMILES or structure files)
• Confidence scoring: output per-pose confidence estimates to prioritize plausible bindings
• Batch docking & ensembles: handle single and multi-complex docking, including protein conformation ensembles
• Use Case: evaluate a library of ligands against a target protein to identify high-confidence poses for experimental follow-up

Quick Start

Run a docking workflow by providing a protein (PDB or sequence) and a ligand (SMILES or structure) to generate pose predictions and confidence scores.