pytdc

What problem does it solve?

This Skill provides a unified interface to access and utilize a vast collection of curated datasets and benchmarks specifically designed for drug discovery and therapeutic machine learning, simplifying complex data acquisition and preprocessing.

Core Features & Use Cases

• Access Diverse Datasets: Load datasets for molecular property prediction (ADME, toxicity), drug-target interactions (DTI), molecular generation, and more.
• Standardized Benchmarks: Utilize pre-defined benchmark groups for systematic model evaluation.
• Data Splitting & Utilities: Employ various splitting strategies (scaffold, cold-split) and chemical utility functions for robust model development.
• Use Case: Train a machine learning model to predict drug toxicity by loading the 'ClinTox' dataset, applying a scaffold split, and using TDC's evaluation metrics to assess performance.

Quick Start

Use the pytdc skill to load the 'Caco2_Wang' ADME dataset and get a scaffold split.