How does PyTDC handle data splitting for ML training?

PyTDC supports advanced splitting strategies such as scaffold splits for chemical diversity and cold-start splits for unseen drugs or targets in interaction tasks.

Is PyTDC suitable for benchmarking new models?

Absolutely. It features Benchmark Groups with curated datasets and specific evaluation protocols (like the 5-seed requirement) for systematic model comparison.

What kind of datasets does the PyTDC skill provide?

It provides AI-ready datasets for ADME, toxicity, drug-target interactions (DTI), drug-drug interactions (DDI), molecular generation, and retrosynthesis.

Can I use this skill to evaluate molecular properties?

Yes, it includes molecular oracles that can score SMILES strings for properties like bioactivity, synthetic accessibility, and drug-likeness.

PyTDC (Therapeutics Data Commons)

Name: PyTDC (Therapeutics Data Commons)
Author: Sologa

bySologa

데이터 과학 및 ML

Accesses AI-ready drug discovery datasets and benchmarks for therapeutic machine learning and pharmacological prediction.

소개

PyTDC is a specialized skill for the Therapeutics Data Commons platform, designed to provide researchers and developers with high-quality, AI-ready datasets for the entire drug discovery pipeline. It offers curated data for single-instance prediction (ADME, toxicity), multi-instance interactions (DTI, DDI), and molecular generation tasks. By integrating standardized evaluation metrics, sophisticated data splits like scaffold-based partitioning, and molecular oracles, this skill enables efficient benchmarking and development of machine learning models within the pharmaceutical and therapeutic domains.

주요 기능

Integrated chemical utilities for format conversion and molecule filtering
Molecular oracles for property scoring and goal-directed generation
Advanced data splitting strategies including scaffold and cold-start splits
Curated access to ADMET, toxicity, and drug-target interaction datasets
0 GitHub stars
Standardized evaluation metrics for classification and regression tasks

사용 사례

Optimizing novel molecule generation using therapeutic oracles
Benchmarking ML models on standardized pharmaceutical datasets
Predicting molecular pharmacokinetic and toxicity properties

소개

주요 기능

Integrated chemical utilities for format conversion and molecule filtering
Molecular oracles for property scoring and goal-directed generation
Advanced data splitting strategies including scaffold and cold-start splits
Curated access to ADMET, toxicity, and drug-target interaction datasets
0 GitHub stars
Standardized evaluation metrics for classification and regression tasks

사용 사례

Optimizing novel molecule generation using therapeutic oracles
Benchmarking ML models on standardized pharmaceutical datasets
Predicting molecular pharmacokinetic and toxicity properties