Chemometric characterization of wines according to their HPTLC fingerprints

Abstract

The objective of this study was to determine which major grape varieties are present in a given wine using both high-performance thin-layer chromatography (HPTLC) fingerprinting and multivariate analysis. For this purpose, 40 mono-and multi-varietal commercial wine samples from four vintages between 2003 and 2012 were collected and analyzed for their polyphenolic composition using HPTLC peak profiles. Polyphenolic compounds such as gallic acid, caffeic acid, resveratrol and rutin (each belonging to one of the four common classes of wine polyphenolic antioxidants) were identified. Unsupervised chemometric method, principal component analysis was used to analyze variance in HPTLC patterns as a function of wine grape variety. An artificial neutral network, as the efficient supervised chemometric tool, was used to develop a predictive model for classification of wine samples and discrimination between them.