QSAR Models for Predicting Aquatic Toxicity of Esters Using Genetic Algorithm-Multiple Linear Regression Methods.

Area of Science:

• Environmental Chemistry
• Toxicology
• Computational Chemistry

Background:

• Esters are vital in various industrial and medical applications.
• Understanding ester toxicity is crucial for environmental and health safety.
• Aquatic toxicity assessment is essential for ecological risk evaluation.

Purpose of the Study:

• To develop a Quantitative Structure-Activity Relationship (QSAR) model for predicting the aquatic toxicity of aliphatic esters.
• To identify key molecular descriptors influencing ester toxicity towards Tetrahymena pyriformis.
• To provide a predictive tool for assessing the toxicity of novel ester derivatives.

Main Methods:

• A dataset of 48 aliphatic esters was utilized, divided into training (34) and test (14) sets.
• Molecular descriptors were calculated using Dragon software.
• Genetic Algorithm (GA) and Multiple Linear Regression (MLR) were employed for descriptor selection and model generation.

Main Results:

• The best QSAR model achieved a high predictive accuracy with R² = 0.899 and Q² LOO = 0.928.
• Leave-One-Out cross-validation and an external test set confirmed the model's predictive power.
• The GA-MLR model identified two significant molecular descriptors correlating structural features with toxicity.

Conclusions:

• The developed GA-MLR QSAR model demonstrates satisfactory predictive ability for ester aquatic toxicity (log 1/IGC50).
• This model can be effectively used for designing similar ester compounds and predicting their toxicity.
• The findings contribute to the safer design and application of ester derivatives in various fields.