Structure-activity relationships for skin sensitization: recent improvements to Derek for Windows

Area of Science:

• Computational toxicology
• Chemical safety assessment
• In silico toxicology

Background:

• Derek for Windows (DfW) is a rule-based expert system for predicting chemical toxicity from structure.
• Predictions rely on alerts identifying structural features (toxicophores) linked to adverse effects.
• Skin sensitization is a key toxicological endpoint requiring accurate predictive models.

Purpose of the Study:

• To report recent enhancements to the skin sensitization alerts in the DfW knowledge base.
• To refine the prediction of skin sensitization potential for specific chemical classes.
• To improve the accuracy and reliability of in silico toxicity assessments.

Main Methods:

• Collaborative effort with Unilever to update DfW knowledge base.
• Modifications to alerts for aldehydes, 1,2-diketones, and isothiazolinones.
• Enhancements included toxicophore definition, mechanistic classification, and supporting evidence.

Main Results:

• Improved alert definitions for aldehydes, 1,2-diketones, and isothiazolinones.
• Refined mechanistic understanding of skin sensitization pathways.
• Strengthened the evidential basis for toxicophore alerts.

Conclusions:

• Updates to DfW significantly enhance the prediction of skin sensitization.
• Continuous refinement of computational models is crucial as new data emerge.
• Collaboration is vital for advancing predictive toxicology tools.