ToxCML: Multi-End Point Toxicity Predictions for 54,601 Chemicals

54,601 unique chemicals with predicted toxicity for 18 end points. The dataset was created by Fauzan Syarif Nursyafi and published on figshare in April 2026. It results from a hybrid computational platform combining consensus QSAR and read-across models.

Use Cases

• Screen new chemicals for potential toxicity based on multi-end point predictions.
• Prioritize chemicals for further testing based on the hybrid mfCoQ-RASAR model scores.
• Compare QSAR and read-across predictive performance for specific toxicity endpoints.
• Analyze chemical space coverage and applicability domain for toxicity models.

Strengths

• Covers 54,601 unique chemicals, providing a broad chemical space.
• Predicts 18 distinct toxicity end points per chemical.
• Models achieved reported AUC scores of approximately 0.86 to 0.99.
• Includes tiered applicability-domain analysis and chemical-space mapping.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data may reflect bias inherent to the training data used for the underlying models.

Provenance

Source

figshare

Collection Method

Generated by the ToxCML computational platform unifying consensus QSAR and similarity-based read-across.

Time Range

The dataset's temporal coverage is not specified.

Freshness

Last updated 2026-04-13 17:37:20; freshness should be verified.

Geography

The dataset's geographic coverage is not specified.