Is a GPU required to use this skill?

While DiffDock can run on a CPU, a GPU is strongly recommended for practical use as it provides a 10-100x speedup, which is essential for virtual screening or batch processing.

Can I use DiffDock without a PDB file?

Yes, DiffDock supports protein sequences via ESMFold integration, allowing you to dock ligands to proteins where only the amino acid sequence is available.

What do the DiffDock confidence scores signify?

Confidence scores represent the model's certainty about the predicted 3D structure. Scores > 0 are high confidence, -1.5 to 0 are moderate, and < -1.5 are low. Note that high confidence in a pose does not necessarily mean high binding affinity.

Does DiffDock predict binding affinity?

No, DiffDock focuses on predicting the physical 3D pose and provides a confidence score for that pose. You should use complementary tools like GNINA or MM/GBSA for binding affinity (delta G) calculations.

DiffDock Molecular Docking

Name: DiffDock Molecular Docking
Author: Activer007

byActiver007

Data Science & ML

Predicts 3D protein-ligand binding poses using state-of-the-art diffusion-based deep learning models for structure-based drug design.

About

DiffDock is a specialized Claude Code skill designed for computational chemists and drug discovery researchers to automate molecular docking workflows. It leverages diffusion models to predict accurate 3D orientations of small molecules within protein targets, providing confidence scores to assess prediction reliability. By supporting both single-complex tasks and high-throughput virtual screening, this skill streamlines the transition from SMILES strings and PDB files to validated structural hypotheses, helping researchers identify how molecules bind to targets without requiring pre-defined binding pockets.

Key Features

Support for protein structures (PDB) and amino acid sequences (ESMFold)
0 GitHub stars
High-accuracy 3D ligand binding pose prediction using diffusion models
Automated batch processing for high-throughput virtual screening campaigns
Flexible input handling for SMILES, SDF, and MOL2 ligand formats
Integrated confidence scoring to quantify prediction reliability

Use Cases

Predicting the binding site and orientation of a novel drug candidate
Screening large compound libraries against a specific protein target
Generating structural inputs for downstream molecular dynamics or affinity scoring

About

Key Features

Support for protein structures (PDB) and amino acid sequences (ESMFold)
0 GitHub stars
High-accuracy 3D ligand binding pose prediction using diffusion models
Automated batch processing for high-throughput virtual screening campaigns
Flexible input handling for SMILES, SDF, and MOL2 ligand formats
Integrated confidence scoring to quantify prediction reliability

Use Cases

Predicting the binding site and orientation of a novel drug candidate
Screening large compound libraries against a specific protein target
Generating structural inputs for downstream molecular dynamics or affinity scoring