Therapeutics Data Commons: Multimodal Benchmarks for Drug Discovery

Therapeutics Data Commons (TDC) is a collection of multimodal benchmarks and datasets for drug discovery and therapeutic science developed by the Harvard MIMS group. Updated as recently as July 2025, it provides a standardized framework for evaluating machine learning models across the drug development pipeline.

Use Cases

• Predicting ADMET properties for drug candidates to assess safety and efficacy
• Modeling drug-target interactions to identify potential therapeutic leads for specific proteins
• Benchmarking graph neural networks on molecular property prediction tasks using SMILES strings

Strengths

• MIT licensed for open commercial and academic research
• Maintained by Harvard MIMS group with active community updates
• Multimodal coverage across drug discovery, development, and clinical stages

Limitations

• Aggregated from various primary sources which may result in heterogeneous data quality across different benchmarks
• Specific row counts and feature schemas vary significantly between individual sub-datasets

Provenance

Source

Harvard University (MIMS group)

Collection Method

Aggregated and curated from existing biomedical literature and public databases

Freshness

Updated July 2025

Geography

Global