Application of an advanced and wide scope non-target screening workflow with LC-ESI-QTOF-MS and chemometrics for the classification of the Greek olive oil varieties

An optimized and validated LC-ESI-QTOF-MS method with an integrated non-target screening workflow was applied in the investigation of the metabolomic profile of 51 Greek monovarietal extra virgin olive oils (EVOOs) from the varieties: Manaki, Ladoelia, Koroneiki, Amfissis, Chalkidikis and Kolovi. Data processing was carried out with the R language and XCMS package. A local database consisting of 1608 compounds naturally occurring in different organs of Olea Europa L. was compiled in order to accelerate the identification workflow. The preliminary examination of the distribution of EVOOs toward their cultivars was achieved by Principal Component Analysis (PCA). Ant Colony Optimization-Random Forest (ACO-RF) was developed to prioritize over 250 features and to establish a classification tree. Apigenin, vanillic acid, luteolin 7-methyl ether and oleocanthal were suggested as the markers responsible for the classification of Greek EVOOs' cultivars.