Figure 2
From: Discovery of food identity markers by metabolomics and machine learning technology
Principal component analyses (PCA) of non-targeted metabolite profiles of three chemical fractions from chia (n = 12), linseed (n = 8) and sesame seeds (n = 8). Matrices of averaged mass features across technical replicates (t) from the VOC (t = 6), POL (t = 5) and SOL (t = 2–5) profiles of the 28 seed batches (S_01 - S_28) were submitted to PCA analysis (Table S2). The first two components of each separate PCA, VOC (A), POL (B), SOL (C), and a PCA of the combined data sets (D) are plotted. Log10-transformed mean-centred ratios of each mass feature were calculated prior to PCA. Missing value (NA) replacement was an estimate of the detection limit (Fig. 4) before calculation of log10-transformed ratios.
