Foodinformatics: Quantitative Structure-Property Relationship Modeling of Volatile Organic Compounds in Peppers

Area of Science:

• Food informatics and chemoinformatics
• Analytical chemistry and chromatography
• Computational chemistry and QSPR modeling

Background:

• Volatile organic compounds (VOCs) significantly impact food aroma and quality.
• Accurate identification and prediction of VOCs are crucial in food analysis.
• Existing methods require robust predictive models for complex food matrices like peppers.

Purpose of the Study:

• To develop a food informatics (chemoinformatics) quantitative structure-property relationship (QSPR) model for pepper VOCs.
• To predict retention indices of VOCs using molecular descriptors and fingerprints.
• To create a stable and predictive model applicable to food analysis.

Main Methods:

• Comprehensive two-dimensional gas chromatography coupled with quadrupole-mass spectrometry (GC × GC/qMS) to measure retention indices of 273 VOCs.
• Calculation of conformation-independent molecular descriptors and fingerprints using Dragon and PaDEL-Descriptor software.
• Application of Balanced Subsets Method (BSM) for dataset division, V-WSP for variable reduction, and Replacement Method (RM) for supervised selection, resulting in a four-descriptor model.

Main Results:

• A stable and predictive QSPR model was established with negligible differences in performance across training, validation, and test sets.
• The model demonstrated high quality, indicated by strong coefficients of determination and low root-mean-square deviations.
• The developed model adheres to OECD principles for applicability in scientific research.

Conclusions:

• The QSPR model provides a reliable mathematical tool for predicting VOC retention indices in peppers.
• This predictive model can assist chromatographers in identifying unknown VOCs in peppers and other food products.
• The study highlights the utility of GC × GC/qMS with bi-dimensional stationary phases for food aroma analysis.