Aggregate entropy scoring for quantifying activity across endpoints with irregular correlation structure.

Area of Science:

• Computational toxicology
• Systems biology
• Environmental health

Background:

• Characterizing systems-level toxicological responses to chemical perturbations is crucial but challenging with diverse in vivo phenotypic data.
• Existing univariate methods struggle to capture complex, multi-endpoint responses and can be confounded by irregular data correlations.

Purpose of the Study:

• To introduce and validate Aggregate Entropy (AggE), an information-theoretic method for scoring multiplexed phenotypic endpoints.
• To assess AggE's performance in identifying chemical effects using developing zebrafish (Danio rerio) across a wide concentration range.

Main Methods:

• Application of the Aggregate Entropy (AggE) method to analyze high-dimensional phenotypic data from zebrafish exposed to 1060 diverse chemicals.
• Evaluation of AggE's accuracy in identifying significant morphological effects, including single and multi-endpoint responses.
• Testing AggE on simulated and real-world datasets to characterize its performance and optimal implementation parameters.

Main Results:

• AggE successfully identified chemicals with significant morphological effects, surpassing univariate methods in detecting multi-endpoint responses.
• The method demonstrated robustness in avoiding false-positives often caused by complex correlation structures in phenotypic data.
• Performance characterization across various datasets provided guidance for practical application.

Conclusions:

• Aggregate Entropy (AggE) offers a robust computational approach for analyzing complex toxicological data from multiplexed phenotypic endpoints.
• AggE enhances the accuracy of chemical hazard identification in environmental and clinical toxicology.
• The method is adaptable for diverse experimental scenarios and high-dimensional biological data analysis.