How can structural similarity analysis help in category formation?

Area of Science:

• Toxicology and Chemical Safety
• Computational Chemistry
• Regulatory Science

Background:

• Chemical categorization is crucial for regulatory programs (e.g., OECD, EU REACH) to address safety data gaps and reduce animal testing.
• Existing categorization approaches aim to streamline chemical management and safety evaluations.
• Computational methods offer potential for enhancing the efficiency and accuracy of chemical grouping.

Purpose of the Study:

• To evaluate the feasibility of using computational structural similarity methods to improve chemical category formation.
• To explore the translation of mechanistic understanding into computer-readable formats for practical applications.
• To propose a novel computational method for augmenting chemical categorization based on reactivity.

Main Methods:

• Utilized computational structural similarity methods to analyze chemical properties.
• Focused on the skin sensitization endpoint as a case study.
• Developed and proposed a novel method called atom environments ranking (AER) for similarity assessment.

Main Results:

• Demonstrated that mechanistic understanding of chemical reactivity can be leveraged by computational methods for category formation.
• Showcased the potential of AER to assess similarity to reference training sets representing common mechanisms of action.
• Indicated feasibility of using computational approaches to group chemicals into distinct reactivity domains.

Conclusions:

• Computational structural similarity methods are feasible for augmenting chemical category formation, particularly for endpoints like skin sensitization.
• The proposed AER method provides a way to group chemicals based on shared mechanisms of action and reactivity.
• Translating mechanistic insights into computational formats is essential for the broad, practical application of chemical categorization.