Putative assignments of selected 12
C metabolites from HCA clusters in Figure 2
. The corresponding cluster numbers from Figure 2 and the mean log2 ratios relative to control for each metabolite are shown along the left side. Orange diamonds = PC patients, grey circles = control subjects. The horizontal whiskers represent the 25th to 75th percentiles of the control-normalized ratios. Adjacent columns in order from left to right indicate the detected accurate masses, the computationally predicted molecular formulas, the putative identities, and the mass spectrometry source modes. Cluster 1 represented a diverse group of metabolites increased in PC patients relative to controls, while clusters 2, 3 and 4 represented reductions in plasmalogen ethanolamines (PlsEtns), phosphocholine-containing metabolites (including PtdCho, lysoPC and SMs), and ultra long-chain fatty acids, respectively. Predicted side-chain speciations in the case of glycerolipids are shown in brackets. Bolded masses with asterisks indicate those selected by Random Forest classification.