alterlab-diffdock

This workflow predicts protein-ligand binding poses by accepting a protein PDB or sequence and a ligand SMILES string, then executing a diffusion-model docking workflow to output candidate 3D poses and confidence scores.

Yes, batch docking handles protein-ligand pose prediction for multiple complexes. You can evaluate a library of ligands against a target protein to identify high-confidence binding poses for experimental follow-up.

Yes, molecular docking with this diffusion model requires a configured environment with Python, PyTorch, RDKit, and DiffDock checkpoints installed to execute the protein-ligand binding pose prediction workflow.

Diffusion-based docking is a method that predicts 3D protein-ligand binding poses by applying a diffusion model to generate candidate conformations, outputting per-pose confidence scores to prioritize plausible bindings.

Confidence scoring for molecular docking poses outputs per-pose estimates to prioritize plausible bindings, allowing you to filter candidate protein-ligand complexes generated by the diffusion model for experimental follow-up.

Yes, structure-based docking predictions support protein conformation ensembles. The diffusion-model workflow handles single and multi-complex docking across PDBs, SMILES, and protein sequences generated via ESMFold.