Tox21 10K Chemical Library with Predicted Properties and Assay Data

The Tox21 10K Compound Library is a U.S. EPA-curated collection of chemicals for toxicology research. It includes predicted physicochemical properties, toxicities, and binarized assay hit calls from ToxCast and Tox21 high-throughput screening. The dataset supports cheminformatics and chemical safety assessment.

Use Cases

• Train QSAR models using predicted physicochemical properties and toxicities from OPERA and T.E.S.T. models.
• Analyze chemical space and diversity using ToxPrint fingerprints (V2.0_r711) for chemotype enrichment.
• Link chemical identifiers by mapping Tox21 IDs to NCGC, PubChem, and DSSTox IDs across partner libraries.
• Investigate structure-activity relationships using binarized assay hit call matrices from invitroDBv2 and invitroDBv3.

Strengths

• Curated by the U.S. Environmental Protection Agency and associated with a peer-reviewed publication in Chemical Research in Toxicology.
• Includes multiple data tables covering chemical identifiers, structures, predicted properties, and experimental assay results.
• Provides QSAR-ready SMILES and standardized identifiers (DSSTox IDs) for interoperability.

Limitations

• The dataset's specific row counts and column counts are not provided, limiting precise assessment of scale.
• Data tables are timestamped from early 2020, which may not reflect the most recent chemical safety assays or models.
• Relies on predicted properties from models like OPERA and Derek Nexus, which carry inherent model uncertainty.

Provenance

Source

U.S. Environmental Protection Agency (EPA).

Collection Method

Compiled from the Tox21 collaborative program, incorporating data from NCATS, NTP, EPA, and predictions from computational models.

Time Range

Data tables are snapshot from early 2020.

Freshness

Data tables are dated January-February 2020; the associated publication is from 2021.

Geography

Primarily United States, as part of the U.S. federal Tox21 program.