Classification of chemicals based on concentration-dependent toxicological data using ToxClust

Area of Science:

• Environmental toxicology
• Computational biology
• Systems toxicology

Background:

• Concentration-response relationships are vital for understanding chemical toxicity.
• High-throughput technologies generate large datasets, posing analysis challenges for environmental scientists.
• Efficient data management and interpretation are crucial for toxicological studies.

Purpose of the Study:

• To introduce ToxClust, a novel method for clustering chemicals based on concentration-response data.
• To provide an effective tool for analyzing and visualizing complex toxicological datasets.
• To facilitate hypothesis generation and mechanism-based chemical classification.

Main Methods:

• ToxClust utilizes a least distance-searching algorithm (LDSA).
• It measures pattern dissimilarity of concentration-response curves between chemicals.
• The method analyzes data from single or multiple toxicological endpoints.

Main Results:

• ToxClust successfully clustered simulated data based on response patterns.
• The method effectively grouped five model chemicals from an in vitro steroidogenesis assay.
• Dissimilarities in concentration-response curves were used for chemical separation.

Conclusions:

• ToxClust is an effective tool for analyzing and visualizing toxicological concentration-response data.
• The method aids in identifying chemical similarities and differences.
• ToxClust supports hypothesis generation and mechanism-based chemical classification.