Predict protein-molecule binding poses

Predict protein-ligand binding poses via diffusion models

Structure-based drug discovery when you need 3D binding poses and pose rankings for virtual screening or lead optimization.

• · PDB file or protein sequence
• · ligand SMILES/SDF/MOL2 file
• · optional protein embeddings

• · ranked SDF poses with confidence scores
• · docking result CSV

• · Python 3.9+
• · PyTorch/PyG
• · RDKit
• · optional CUDA GPU

DiffDock v1.1.3 installed via conda/Docker, GPU recommended for speed

Predicts poses only, not binding affinity; needs external scoring (GNINA/MM-GBSA) for ΔG; confidence scores may be overconfident on novel scaffolds

• · State-of-the-art diffusion-based docking
• · Batch processing for virtual screening
• · Confidence scoring integration