Prioritizing steatogenic chemicals through integration ToxCast™ data, machine learning, and experimental validation.

Area of Science:

• Toxicology
• Environmental Health
• Metabolic Disease Research

Background:

• Hepatic steatosis (fatty liver disease) is a growing global health concern linked to environmental exposures.
• Effective risk-oriented regulation requires identifying and prioritizing chemicals with high steatogenic potency.

Purpose of the Study:

• To develop and validate a predictive model for identifying high-potency steatogenic chemicals.
• To integrate high-throughput screening data with in vivo and in vitro experimental validation.

Main Methods:

• Leveraged the adverse outcome pathway framework for hepatic steatosis.
• Integrated ToxPi scores from ToxCast™ database with zebrafish in vivo validation.
• Employed Support Vector Machine (SVM) and Random Walk with Restart algorithms for prediction and categorization.

Main Results:

• Developed a SVM model with 91.7% accuracy on the training set and 77.1% on external validation.
• Categorized 345 curated chemicals: 37.97% high-potency, 18.84% moderate-potency, 43.19% null-effect.
• Confirmed high steatogenic potency for emerging contaminants like isodecyl diphenyl phosphate and tetrabromobisphenol A bis(2-hydroxyethyl) ether.

Conclusions:

• The integrated approach effectively prioritizes high-potent steatogenic chemicals.
• Findings highlight the risk posed by several emerging environmental contaminants.
• This methodology supports robust toxicological and environmental risk assessment.