English: A) Heat map of diverse Japanese green tea extracts. Columns represent metabolic profile of single cultivars, rows represent 541 individual analytes. Colors correspond to relative metabolite areas. B) PCA score plot shows separate clustering of MS profiles corresponding to Nou-6 and SR, and other cultivars. C) Corresponding loading plots of all samples show MS peaks that differ among samples. D) PCA score plot derived from three representative tea cultivars; Yabukita (YB), Benifuuki (BF) and SR. E) Corresponding loading plots of three tea cultivars (YB, BF, and SR) show MS peaks that differ among samples. OPLS-DA score plots (F and G) and loading S-plots (H and I) were derived from each LC-MS data set (F and H: YB vs SR; G and I: BF vs SR). S-plot shows covariance w against correlation p (corr) of variables of the discriminating component of OPLS-DA model. Cut-off values for p (corr) < |0.8| were used to select metabolites that most strongly contributed to differences between two tea extracts.