Predicted NOEC Toxicity for 10,863 PFAS Chemicals Across 13 Fish Species

Machine learning-predicted no-observed-effect concentrations (NOECs) for 10,863 per- and polyfluoroalkyl substances (PFAS) in Danio rerio (zebrafish), with extrapolations to 12 additional fish species. The model uses compiled toxicity data and molecular structural descriptors, achieving a test R² of 0.7096. The dataset was created by Minhao Wang and last updated on 2026-05-21.

Use Cases

• Prioritizing PFAS chemicals for regulatory review based on predicted toxicity ranges from 0.239 to 163.452 mg/L.
• Investigating the 'U-shaped' toxicity chain-length trend and the influence of polar functional groups like sulfonates and carboxylates.
• Assessing trait-mediated differences in toxicity sensitivity across 13 fish species, including warm water and omnivorous species.
• Screening ecological risks by integrating chemical structure descriptors with species ecological characteristics.

Strengths

• Covers a large chemical space with predictions for 10,863 PFAS substances.
• Model performance is quantified with a test R² value of 0.7096.
• Extends predictions beyond a single model organism to 13 total fish species via trait-based inference.
• Provides specific predicted NOEC ranges, such as 0.239 (0.131–1.129) to 163.452 (77.207–341.332) mg/L.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Predictions are based on a machine learning model; experimental validation for all chemicals is not provided.

Provenance

Source

figshare

Collection Method

Machine learning model trained on compiled toxicity data using molecular structural descriptors.

Freshness

Last updated 2026-05-21 09:22:35; freshness should be verified.