Hierarchical Mechanistic Modeling of Complex Toxicity Endpoints from Public Concentration-Response Data.

Area of Science:

• Computational toxicology
• Chemical safety assessment
• Adverse outcome pathway (AOP) research

Background:

• High-throughput screening (HTS) generates vast chemical toxicity data.
• Interpreting HTS data for predictive modeling is challenging due to data inconsistencies and varied experimental designs.

Purpose of the Study:

• To develop a hierarchical mechanistic modeling framework for structuring and interpreting HTS concentration-response data.
• To create AOP-based models linking chemical bioactivity to adverse outcomes.

Main Methods:

• Integrated curated data from 455 PubChem assays, mapping to 216 protein targets and 103 WikiPathways.
• Organized assay data in a biologically layered hierarchy to build AOP-based models.
• Generated pathway-level toxicity scores by integrating protein activity and pathway perturbations.

Main Results:

• Developed a framework linking chemical bioactivity to 5 in vivo toxicity endpoints (acute systemic, maternal, developmental, hepatotoxicity).
• 103 pathways were statistically associated with these toxicity endpoints.
• The models quantified compound potency and predicted diverse toxicity outcomes.

Conclusions:

• The hierarchical framework enhances mechanistic interpretability of pathway-level effects from HTS data.
• Provides a quantitative basis for compound ranking, potency assessment, and hazard prediction.
• Supports computational toxicology, chemical risk assessment, and early-stage drug discovery.