Comparison of ready biodegradation estimation methods for fragrance materials

Area of Science:

• Environmental Chemistry
• Computational Chemistry
• Green Chemistry

Background:

• Biodegradability assessment is crucial for evaluating environmental exposure and risks of organic chemicals.
• Predictive models facilitate regulatory compliance and green chemistry initiatives by offering accessible tools.

Purpose of the Study:

• To evaluate the performance of free, readily available estimation programs for predicting the ready biodegradability of fragrance materials.
• To compare 10 different models using a diverse dataset of 222 fragrance compounds.

Main Methods:

• A database of 222 fragrance compounds with measured ready biodegradability data was utilized.
• Ten predictive models, including VEGA, START, Biowin (1-6), and two machine learning models, were compared.
• Performance metrics included overall accuracy, sensitivity, specificity, and Matthews correlation coefficient (MCC). Applicability Domain (AD) was also assessed.

Main Results:

• Overall accuracy across models was approximately 70%, with minimal variation.
• Sensitivity, specificity, and MCC showed greater differences between models.
• Biowin6, VEGA, and Biowin3 demonstrated the highest predictive quality based on MCC.
• VEGA showed improved reliability for predictions within its applicability domain (AD index >0.8).
• Excluding compounds with quaternary carbons improved predictivity for VEGA, START, and Biowin3/6.

Conclusions:

• VEGA, Biowin3, and Biowin6 are suitable for screening the ready biodegradability of fragrance compounds.
• These models offer valuable tools for environmental risk assessment and the design of greener chemicals.