Classification of Different Dried Vine Fruit Varieties in China by HS-SPME-GC-MS Combined with Chemometrics

The volatile profiles of six dried vine fruit varieties produced in Xinjiang province were detected by HS-SPME–GC-MS. In total, 80 compounds were identified; among them, rose oxide, linalool oxide A, linalool oxide B, 3,7-dimethyl-1,5-octadien-3,7-diol, 2-heptenol, and 3-octen-2-one are reported for the first time in dried vine fruits. Chemometrics was used to classify the different dried vine fruit varieties. Principal component analysis (PCA) overviewed the data structure of the six dried vine fruit varieties and two subsets—muscat and non-muscat varieties obtained. Partial least squares discriminant analysis (PLS-DA) was conducted to classify different varieties based on their volatile compounds and generated a perfect classification model with an accuracy of 100% for the external test set. Additionally, to find out potential markers, two variable selection methods were combined with the PLS-DA model and compared—variable importance in projection (VIP) and sensitivity ratio (SR). The SR method was more efficient at identifying the potential volatile markers. The potential markers can successfully classify different dried vine fruit varieties.