Is a GPU required to run this skill efficiently?

While it can run on a CPU, a GPU is strongly recommended. Using a GPU provides a 10-100x speedup compared to a CPU, which is essential for virtual screening or batch processing.

Can I use DiffDock if I only have a protein sequence?

Yes, DiffDock can utilize ESMFold to predict the protein structure from an amino acid sequence before performing the docking task.

What types of ligands are supported by the DiffDock skill?

DiffDock supports common chemical formats including SMILES strings, SDF files, and MOL2 files, primarily focused on small molecule organic compounds between 100-1000 Da.

Does DiffDock predict binding affinity?

No, DiffDock focuses on predicting the 3D binding pose (structure) and a confidence score. For binding affinity (ΔG) or ranking by potency, results should be combined with specialized scoring functions like GNINA or MM/GBSA.

DiffDock Molecular Docking

Name: DiffDock Molecular Docking
Author: Zehong-Wang

byZehong-Wang

0•

数据科学与机器学习

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

DiffDock is a specialized Claude Code skill designed for computational chemists and bioinformaticians to perform high-accuracy molecular docking. By leveraging diffusion-based deep learning, it predicts the 3D structures of small molecule ligands bound to protein targets, supporting both PDB files and amino acid sequences via ESMFold. This skill streamlines complex workflows for structure-based drug design, lead optimization, and virtual screening by providing automated docking pipelines, confidence scoring for prediction reliability, and efficient batch processing capabilities for large compound libraries.

主要功能

01High-accuracy 3D binding pose prediction using diffusion models

02Support for protein structures (PDB) and amino acid sequences (ESMFold)

03Cross-format support for SMILES, SDF, and MOL2 ligand inputs

04Reliability assessment via deep learning-based confidence scores

05Automated batch processing for large-scale virtual screening

060 GitHub stars

使用场景

01Predicting how a drug candidate binds to a specific protein target

02Generating initial binding poses for downstream molecular dynamics or affinity scoring

03Running virtual screening on chemical libraries to identify potential leads

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add zehong-wang/kosmos diffdock

For use in Claude.ai and ChatGPT

Download Skill