PyTDC (Therapeutics Data Commons) is a platform that provides curated, AI-ready datasets and benchmarks specifically for drug discovery and therapeutic development.

Does it support protein-related tasks?

Yes, it includes datasets for drug-target interactions (DTI), protein-protein interactions (PPI), and epitope prediction for biologics.

What data splitting methods does it support?

It supports standard random splits as well as domain-specific splits like scaffold splits for chemical diversity and 'cold' splits for testing generalization to unseen drugs or targets.

What are molecular oracles?

Oracles are functions provided by PyTDC to evaluate specific properties of molecules (like binding affinity or drug-likeness), which are used to guide molecule generation models.

Can this skill help with ADMET prediction?

Yes, it provides datasets for Absorption, Distribution, Metabolism, Excretion, and Toxicity (ADMET) properties used to evaluate the safety and efficacy of drug candidates.

Therapeutics Data Commons (PyTDC)

Name: Therapeutics Data Commons (PyTDC)
Author: jimmc414

byjimmc414

•

324

데이터 과학 및 ML

Provides AI-ready datasets and benchmarks for therapeutic machine learning and drug discovery tasks.

소개

The PyTDC skill equips Claude with specialized capabilities for pharmacological research and drug development. It provides access to the Therapeutics Data Commons, an open-science platform containing curated datasets spanning the entire therapeutics pipeline, including ADME, toxicity, and drug-target interactions. This skill is essential for researchers building machine learning models for molecule generation, property prediction, and benchmarking. It includes support for standardized evaluation metrics, advanced data splitting strategies like scaffold and cold-start, and molecular oracles for property optimization, enabling autonomous discovery workflows within Claude Code.

주요 기능

Curated datasets for ADME, toxicity, and bioactivity prediction
Comprehensive utilities for molecular format conversion and data processing
Standardized benchmark groups for systematic model evaluation
Advanced data splitting strategies including scaffold and cold-splits
324 GitHub stars
Molecular oracles for property optimization and molecule generation

사용 사례

Generating novel lead molecules with optimized chemical properties
Predicting pharmacokinetic and toxicity properties of drug candidates
Benchmarking machine learning models on standardized pharmaceutical tasks

소개

주요 기능

Curated datasets for ADME, toxicity, and bioactivity prediction
Comprehensive utilities for molecular format conversion and data processing
Standardized benchmark groups for systematic model evaluation
Advanced data splitting strategies including scaffold and cold-splits
324 GitHub stars
Molecular oracles for property optimization and molecule generation

사용 사례

Generating novel lead molecules with optimized chemical properties
Predicting pharmacokinetic and toxicity properties of drug candidates
Benchmarking machine learning models on standardized pharmaceutical tasks