Does TorchDrug work with protein structures?

Yes, it supports protein modeling using both sequences and 3D structures (PDB files), including specialized models like GearNet and ESM.

What graph architectures does this skill support?

It provides implementation patterns for over 20 architectures, including GIN, GAT, SchNet, MPNN, and knowledge graph embedding models like RotatE.

Is TorchDrug compatible with RDKit?

Absolutely. This skill provides integration patterns to convert between TorchDrug objects and RDKit molecules for advanced cheminformatics tasks.

What is the TorchDrug Claude Code skill?

It is a specialized capability that provides Claude with the knowledge to implement drug discovery workflows using the TorchDrug library, covering molecules, proteins, and knowledge graphs.

Can I use this skill for chemical retrosynthesis?

Yes, the skill includes guidance for predicting synthetic routes and identifying reaction centers using models trained on datasets like USPTO-50k.

TorchDrug Molecular Science

Name: TorchDrug Molecular Science
Author: Zehong-Wang

byZehong-Wang

0•

Data Science & ML

Accelerates drug discovery and molecular research using graph neural networks and PyTorch-based machine learning.

TorchDrug is a specialized Claude Code skill that empowers researchers and developers to perform advanced drug discovery tasks directly within their coding environment. By providing domain-specific guidance on molecular property prediction, protein modeling, and knowledge graph reasoning, it streamlines the implementation of complex Graph Neural Network (GNN) architectures like GIN and SchNet. This skill is essential for building AI-driven pipelines for ADMET prediction, de novo molecular generation, or retrosynthesis planning using TorchDrug's extensive catalog of 40+ curated datasets and deep integration with the PyTorch ecosystem.

Key Features

01Automated retrosynthesis planning and chemical reaction center identification

02Generative models for de novo molecular design and lead optimization

03Biomedical knowledge graph reasoning for drug repurposing and link prediction

040 GitHub stars

05Comprehensive molecular property prediction for ADMET and toxicity screening

06Advanced protein modeling using sequence-based and structure-aware GNNs

Use Cases

01Analyzing protein-protein interactions and enzyme functions using structural data

02Mining biomedical knowledge graphs to discover new gene-disease associations

03Predicting the solubility, toxicity, and biological activity of novel drug candidates

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add zehong-wang/kosmos torchdrug

For use in Claude.ai and ChatGPT

Download Skill