Predict 3D protein-ligand binding poses and confidence scores with diffusion models.
npx skills add https://github.com/rubensliv/k-dense-ai --skill diffdock-rubensliv
Please help me install this Agent Skill. Skill: diffdock Source: https://github.com/rubensliv/k-dense-ai/tree/main/scientific-skills/diffdock Command: npx skills add https://github.com/rubensliv/k-dense-ai --skill diffdock-rubensliv
DiffDock addresses the challenge of generating accurate 3D protein-ligand binding poses along with model-generated confidence scores, streamlining structure-based docking workflows and reducing manual pose exploration.
Run the inference module with your protein input (PDB or sequence) and a ligand (SMILES or file) to generate poses and confidence scores.
High-intent search queries and answers about installing and using this skill.
Predict protein-ligand binding poses by running DiffDock inference with a protein PDB file or sequence and a ligand SMILES or SDF/MOL2 file. The workflow generates 3D pose files and per-pose confidence scores for structure-based docking analysis.
Yes, batch virtual screening is supported using protein sequences via ESMFold and ligand inputs as SMILES. The workflow processes multiple complexes to generate 3D poses and confidence scores suitable for downstream analysis.
Rescore DiffDock poses by feeding the generated 3D pose files and protein structures into GNINA, MM/GBSA, or open-source energy calculation tools. This downstream rescoring validates the predicted ligand poses using independent scoring functions.
Yes, DiffDock supports protein inputs as sequences by utilizing ESMFold to generate the required protein structures. This allows molecular docking workflows when only the protein sequence is available instead of a PDB file.
Yes, diffusion-based molecular docking requires PyTorch Lightning, RDKit, BioPython, and Pandas to operate. These dependencies handle the neural network inference, molecular parsing, and structural data processing.
DiffDock confidence scores provide model-generated estimates for pose prediction reliability but are not absolute binding affinities. Downstream rescoring with GNINA or MM/GBSA is recommended to rigorously evaluate virtual screening hits.
felixboehmPredict protein-ligand binding poses and confidence scores with diffusion models.
OwnLabAIGenerate protein-ligand docking poses with diffusion models and confidence scores.
sanand0Predict 3D binding poses of small molecules to protein targets using diffusion models.
jimmc414Predicts protein-ligand binding poses and confidence scores via molecular docking.
