Figure 3

Multivariate statistical analysis of metabolic profiles derived from various tea extracts. (A) Heat map of 42 green tea cultivars. Columns represent the metabolic profile of single cultivars and rows represent 279 detected peaks in LC–MS measurement. (B) PCA score plot shows separate clustering of MS profiles corresponding to Nou 6 (No.41) and Sunrouge (No.42), and other cultivars. The bioactivity-prediction OPLS model was calculated from the LC–MS dataset of (C) 41 (α-glucosidase) or (D) 27 (α−amylase) tea samples, excluding cultivars that showed an inhibitory rate of 0%. R2, the goodness-of-fit parameter; RMSEE, the root mean squared error of the estimation; RMSEcv, the root mean squared error of cross validation.