Functional recognition of structure-diverse odor molecules in drinking water based on QSOR study

Area of Science:

• Environmental Chemistry
• Water Quality Assessment
• Predictive Modeling

Background:

• Taste and odor issues in drinking water are persistent challenges for water utilities and consumers.
• Identifying specific odor-causing compounds in water remains a significant hurdle for the water industry.

Purpose of the Study:

• To develop a predictive model for odor thresholds of drinking water contaminants.
• To establish a quantitative structure odor relationship (QSOR) for odorant identification.
• To provide a tool for screening potential odor-causing chemicals in water.

Main Methods:

• A linear quantitative structure odor relationship (QSOR) model was built using the partial least squares (PLS) method.
• 22 reported odor compounds with similar characteristics were used as a training set.
• The logarithm of odor threshold (OT) divided by molecular weight (pOT) served as the response descriptor.

Main Results:

• The developed QSOR model achieved strong statistical performance: R²=0.8988, RMSE=0.4374, XR²=0.8133, XRMSE=0.5993.
• External validation using a nonlinear binary QSOR method showed stable accuracy around 90%.
• The model successfully predicted pOT values for 8 test compounds, correlating well with experimental data.

Conclusions:

• The validated QSOR model offers a reliable method for predicting odor thresholds of potential odorants.
• This approach provides a novel and convenient way to screen for odor-causing compounds from a vast number of chemicals.
• The findings can assist water utilities in proactively managing and mitigating drinking water taste and odor problems.