What is TorchDrug used for?

TorchDrug is a comprehensive machine learning toolbox designed for drug discovery, supporting tasks like molecular property prediction, protein modeling, and knowledge graph reasoning.

How does it handle chemical reactions?

It includes a dedicated retrosynthesis pipeline that identifies reaction centers and completes synthons to plan synthetic routes for target molecules.

Can this skill help with ADMET prediction?

Yes, it provides specialized tasks and models for predicting ADMET properties, including toxicity, solubility, and metabolic stability.

Does it support protein structure files?

Yes, TorchDrug can process PDB files and work with 3D protein structures using geometric models like GearNet and SchNet.

TorchDrug Science Toolkit

Name: TorchDrug Science Toolkit
Author: ricable

byricable

•

Ciencia de Datos y ML

Provides an advanced machine learning framework for drug discovery, molecular property prediction, and protein modeling using PyTorch.

Acerca de

The TorchDrug skill equips Claude with specialized knowledge for scientific computing in the pharmaceutical and biochemical domains. It leverages a graph-based framework to perform complex tasks such as molecular property prediction (ADMET), protein sequence and structure analysis, and biomedical knowledge graph reasoning. By integrating over 40 curated datasets and numerous GNN architectures like GIN and SchNet, this skill enables researchers and developers to accelerate drug discovery workflows, ranging from generating novel molecular structures to planning multi-step retrosynthesis routes with production-grade precision.

Características Principales

1 GitHub stars
De novo molecular generation and multi-objective property optimization
Graph-based molecular property prediction for ADMET and toxicity screening
Protein sequence and 3D structure modeling with GearNet and ESM integration
Biomedical knowledge graph reasoning for drug repurposing and link prediction
Automated chemical retrosynthesis planning and reaction center identification

Casos de Uso

Predicting the biological activity and toxicity of drug candidates
Generating novel therapeutic compounds with specific chemical constraints
Analyzing protein functions and structural properties for enzyme engineering

Acerca de

Características Principales

1 GitHub stars
De novo molecular generation and multi-objective property optimization
Graph-based molecular property prediction for ADMET and toxicity screening
Protein sequence and 3D structure modeling with GearNet and ESM integration
Biomedical knowledge graph reasoning for drug repurposing and link prediction
Automated chemical retrosynthesis planning and reaction center identification

Casos de Uso

Predicting the biological activity and toxicity of drug candidates
Generating novel therapeutic compounds with specific chemical constraints
Analyzing protein functions and structural properties for enzyme engineering