Can I use protein sequences instead of PDB files?

Yes, DiffDock supports protein sequences by utilizing ESMFold to generate structure predictions internally before performing the docking process.

Can I use DiffDock for protein-protein docking?

DiffDock is optimized for small molecule ligands (typically 100-1000 Da). For docking two proteins or large peptides, specialized tools like DiffDock-PP or AlphaFold-Multimer are more appropriate.

Is a GPU required to run this skill?

While it can run on a CPU, a GPU is strongly recommended as it provides a 10-100x speedup, which is necessary for practical research and batch screening campaigns.

What ligand formats are supported?

DiffDock is versatile and accepts common chemical formats including SMILES strings, SDF files, and MOL2 files.

Does DiffDock predict binding affinity?

No, DiffDock specifically predicts the 3D binding pose and a confidence score for that pose. For binding affinity (ΔG or Kd), you should use complementary scoring functions like GNINA or MM/GBSA after generating the poses.

DiffDock Molecular Docking

Name: DiffDock Molecular Docking
Author: ricable

byricable

•

データサイエンスとML

Predicts high-accuracy 3D protein-ligand binding poses using diffusion-based deep learning for structure-based drug design.

DiffDock is a state-of-the-art computational biology tool designed to predict the 3D orientation (binding pose) of small molecule ligands within protein targets. By leveraging diffusion models, it provides high-accuracy structural predictions that are essential for virtual screening, lead optimization, and chemical biology research. While it focuses on structural configuration and prediction confidence rather than direct binding affinity, it integrates seamlessly into drug discovery pipelines by supporting batch processing, protein sequences via ESMFold, and compatibility with downstream scoring functions like GNINA.

主な機能

01Support for protein PDB files or raw amino acid sequences via ESMFold

02Flexible ligand input handling including SMILES, SDF, and MOL2

031 GitHub stars

04High-accuracy 3D binding pose prediction using diffusion models

05Batch processing for large-scale virtual screening campaigns

06Reliability assessment through automated confidence scoring

ユースケース

01Virtual screening of large compound libraries against specific receptors

02Structure-based drug discovery and lead compound optimization

03Identifying potential binding sites for novel protein targets

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add ricable/ultimate-ai-agent diffdock

For use in Claude.ai and ChatGPT

Download Skill