Profiling volatile organic compounds of different grape seed oil genotypes using gas chromatography-mass spectrometry and chemometric methods

This study utilized gas chromatography-mass spectrometry (GC-MS) for untargeted metabolomic profiling of grape seed oils (GSO) taken from five major grape genotypes in Iran. A total of 175 volatile organic compounds (VOCs) were identified in the GSO, with 20 identified as core molecules being present in all genotypes and samples, and 155 identified as accessory and rare molecules, found in ≥10 % but <100 % of the samples. We hypothesized that specific VOCs in GSO genotypes could be used as reliable indicators to differentiate genotypes and assess their quality. The core molecules mainly consisted of hydrocarbons (35 %), fatty acids (30 %), aldehydes (15 %), and esters (5 %), with putative names assigned to 7 compounds and putative formulas to 10. Of the 155 accessory and rare molecules, 12 volatile compounds were uniquely identified in distinct GSO genotypes, indicating specific phenotypic characteristics associated with different GSO genotypes. Among 20 core molecules, ten were consistently ranked higher in importance through 70 iterations of the Boruta feature selection algorithm. Fatty acids, including Linoleic and Oleic acid, emerged as key compounds for assessing the quality of the GSO samples. Using 10 core molecules as predictors, supervised learning methods such as random forest, support vector machine, partial least squares discriminant analysis, and k-nearest neighbor achieved 100 % accuracy, sensitivity, specificity, and precision in classifying different GSO genotypes for both training and test sets. The identified metabolites served as potential markers for predicting quality and distinguishing genotypes, highlighting the efficiency of metabolomic profiling in analyzing GSO variations and providing insights into GSO quality.