New Quantitative Structure-Activity Relationship Models Improve Predictability of Ames Mutagenicity for Aromatic Azo...

Area of Science:

• Computational chemistry
• Toxicology
• Medicinal chemistry

Background:

• Existing Quantitative Structure-Activity Relationship (QSAR) models exhibit limitations in predicting the mutagenicity of aromatic azo compounds.
• There is a need for improved predictive models for this specific chemical class.

Purpose of the Study:

• To develop and validate novel QSAR models for predicting Ames mutagenicity in aromatic azo compounds.
• To enhance the accuracy and reduce false positives in mutagenicity predictions.

Main Methods:

• Development of two new QSAR models: one using Simplified Molecular Input Line Entry System (SMILES) descriptors calculated by CORAL software, and another employing the k-nearest neighbors algorithm.
• Evaluation and comparison of the predictive performance of individual models and their combined output.
• Assessment of existing QSAR models for mutagenicity prediction.

Main Results:

• Both newly developed QSAR models demonstrated satisfactory statistical predictive quality for Ames mutagenicity.
• Combining predictions from the two new models led to a significant improvement in overall performance.
• Existing models were effective in identifying some toxic compounds but produced a high rate of false positives.

Conclusions:

• The two novel QSAR models, trained on a larger dataset of related compounds and utilizing improved algorithms, effectively predict aromatic azo compound mutagenicity.
• These models offer a more reliable alternative to existing methods by minimizing false positive predictions.
• The combined approach provides a robust tool for assessing the mutagenic potential of aromatic azo compounds.