Application of an Electronic Nose to the Prediction of Odorant Series in Wines Obtained with Saccharomyces or Non-Saccharomyces Yeast Strains

Abstract

Electronic noses (E-noses) have become powerful tools for the rapid and cost-effective differentiation of wines, providing valuable information for the comprehensive evaluation of aroma patterns. However, they need to be trained and validated using classical analytical techniques, such as gas chromatography coupled with mass spectrometry, which accurately identify the volatile compounds in wine. In this study, five low-ethanol wines with distinctive sensory profiles—produced using Saccharomyces and non-Saccharomyces yeasts and tailored to modern consumer preferences—were analyzed to validate the E-nose. A total of 57 volatile compounds were quantified, 27 of which had an Odor Activity Value (OAV) over 0.2. The content in volatiles, grouped into 11 odorant series according to their odor descriptors, along with the data provided by 12 E-nose sensors, underwent advanced statistical treatments to identify relationships between both data matrices. Partial least squares discriminant analysis (PLS-DA) applied to the data from the 12 E-nose sensors revealed well-defined clustering patterns and produced a model that explained approximately 92% of the observed variability. In addition, a principal component regression (PCR) model was developed to assess the ability of the E-nose to non-destructively predict odorant series in wine. The synergy between the volatile compound profiles and the pattern recognition capability of the E-nose, as captured by PLS-DA, enables a detailed characterization of wine aromas. In addition, predictive models that integrate data from gas chromatography, flame ionization detection, and mass spectrometry (GC-FID/GC-MSD) with the electronic nose demonstrating a promising approach for a rapid and accurate odor series prediction, thereby increasing the efficiency of wine aroma analysis.