diffdock

To predict 3D protein-ligand binding poses using diffusion models, provide a protein structure (PDB or sequence) and a ligand specification (SMILES, SDF, or MOL2) to run diffusion-based docking and generate predicted poses with confidence scores.

Yes, molecular docking accepts PDB files or protein sequences via ESMFold as protein inputs, and SMILES strings or SDF/MOL2 structure files as ligand inputs to predict binding poses.

No, DiffDock does not predict binding affinity; it predicts 3D binding poses for small-molecule ligands and outputs a per-pose confidence score for downstream rescoring workflows.

To handle protein flexibility during virtual screening, use ensemble docking by providing multiple protein structures as inputs, allowing the diffusion model to account for target flexibility when predicting ligand poses.

Yes, batch molecular docking supports CSV handling for multiple ligands, allowing you to screen a compound library against a target to obtain top poses and associated confidence scores.

Yes, rdkit and pandas are required dependencies to run these protein-ligand docking workflows, which include environment checks and parameter templating for processing ligand inputs.