What is the TorchDrug skill for Claude Code?

It is a specialized capability that allows Claude to assist with machine learning tasks in drug discovery, leveraging the TorchDrug library for molecular and protein science.

Which datasets are available through this skill?

It supports over 40 curated datasets, including BBBP, Tox21, QM9, Hetionet, and USPTO-50k for various scientific tasks.

Can this skill help with protein structure prediction?

Yes, it includes models like GearNet and ESM for protein function and structure prediction using sequences or PDB files.

Does TorchDrug support RDKit integration?

Yes, the skill provides patterns for converting between TorchDrug molecular objects and RDKit for advanced cheminformatics workflows.

TorchDrug for Molecular Science

Name: TorchDrug for Molecular Science
Author: jimmc414

byjimmc414

•

324

데이터 과학 및 ML

Empowers Claude to perform graph-based drug discovery, molecular property prediction, and protein modeling using the TorchDrug framework.

소개

TorchDrug is a specialized skill that equips Claude with the tools needed for advanced machine learning in drug discovery and molecular science. Built on PyTorch, it enables high-level tasks such as predicting ADMET properties, analyzing protein structures, reasoning through biomedical knowledge graphs, and generating novel molecular structures. By providing structured access to over 40 curated datasets and dozens of Graph Neural Network (GNN) architectures, this skill allows researchers and developers to automate complex scientific workflows like retrosynthesis planning and lead optimization directly within their AI-driven development environment.

주요 기능

Graph-based molecular property prediction including ADMET, toxicity, and solubility.
De novo molecular generation and multi-objective lead optimization.
Automated retrosynthesis planning and chemical reaction center identification.
Protein modeling capabilities for sequence function and 3D structure prediction.
Biomedical knowledge graph reasoning for drug repurposing and link prediction.
324 GitHub stars

사용 사례

Predicting the solubility and toxicity of drug candidates using GNN architectures like GIN and SchNet.
Discovering new disease treatments via link prediction in large-scale biomedical knowledge graphs like Hetionet.
Designing synthetic routes for complex target molecules using multi-step retrosynthesis pipelines.

소개

주요 기능

Graph-based molecular property prediction including ADMET, toxicity, and solubility.
De novo molecular generation and multi-objective lead optimization.
Automated retrosynthesis planning and chemical reaction center identification.
Protein modeling capabilities for sequence function and 3D structure prediction.
Biomedical knowledge graph reasoning for drug repurposing and link prediction.
324 GitHub stars

사용 사례

Predicting the solubility and toxicity of drug candidates using GNN architectures like GIN and SchNet.
Discovering new disease treatments via link prediction in large-scale biomedical knowledge graphs like Hetionet.
Designing synthetic routes for complex target molecules using multi-step retrosynthesis pipelines.