pytdc

Access standardized drug discovery datasets and molecular evaluation oracles via PyTDC.

2|Updated Jun 4, 2026

Search Tags:#machine learning #bioinformatics #drug discovery #pharmacology #molecular generation #admet

Ready to run

npx skills add https://github.com/Lord1Egypt/scientific-agent-toolkit --skill pytdc-lord1egypt

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: pytdc
Source: https://github.com/Lord1Egypt/scientific-agent-toolkit/tree/main/scientific-skills/pytdc
Command: npx skills add https://github.com/Lord1Egypt/scientific-agent-toolkit --skill pytdc-lord1egypt

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires PyTDC, numpy, pandas, tqdm, seaborn, scikit_learn, fuzzywuzzy, and includes scripts (resource) and references (resource) components.

What problem does it solve?

This skill addresses the fragmentation and lack of standardization in pharmaceutical machine learning by providing a unified interface to access curated drug discovery datasets and evaluation protocols.

Core Features & Use Cases

Standardized Access: Easily load datasets for ADME, toxicity, and drug-target interactions with consistent train/test splits.
Molecular Optimization: Utilize built-in oracle functions to evaluate and optimize novel molecules for specific biological targets.
Benchmarking: Implement rigorous evaluation protocols using standardized metrics and multi-seed testing to ensure model reproducibility.

Quick Start

Use the pytdc skill to load the Caco2_Wang ADME dataset and perform a scaffold split for model training.

Frequently Asked Questions about pytdc

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I load standardized datasets for ADME and toxicity prediction in machine learning?▼

You can load standardized drug discovery datasets for ADME and toxicity prediction using a unified interface that provides consistent train/test splits. This ensures reproducible benchmarking for developing pharmacological machine learning models.

What is the best way to benchmark drug-target interaction models with reproducible splits?▼

The best way to benchmark drug-target interaction models is using standardized evaluation protocols with multi-seed testing and consistent data splits. This approach ensures rigorous model reproducibility and reliable pharmacological metrics.

Can I evaluate and optimize novel molecules for specific biological targets?▼

Yes, you can evaluate and optimize novel molecules for specific biological targets using built-in oracle functions. These molecular evaluation oracles facilitate systematic molecular generation and optimization tasks.

Do I need the PyTDC library to access drug discovery datasets and evaluation oracles?▼

Yes, you need the PyTDC library to manage standardized drug discovery data retrieval, dataset splitting, and benchmark evaluation. It acts as the core dependency for accessing these AI-ready pharmacological datasets.

How does scaffold splitting improve machine learning models for drug discovery?▼

Scaffold splitting improves drug discovery machine learning models by structuring train and test sets based on molecular scaffolds. This standardized data partitioning prevents data leakage and ensures robust model generalization across novel chemical structures.

What limitations exist when using standardized benchmarks for molecular generation?▼

Standardized benchmarks for molecular generation limit customization to predefined dataset splits and oracle evaluation metrics. Users must adhere to these fixed protocols, which may restrict highly specialized pharmacological workflows outside the provided ADME and toxicity tasks.