Fig. 2

Principal component (PC) curves as biochemical indicators. a Loading curves of the first three PCs calculated by PCA. Since PCA is performed on the overall dataset of all probed cells, the computed PCs are the same throughout all of the measured spectra, and their characteristic bands indicate sensitive biochemical features (see text for details concerning peak assignment). While the PC1 curve only resembles the global average spectrum, the PC2 and PC3 curves account for the significant biochemical differences between the different cells, as well as different regions of the same cell. Scatter plot of PC3 vs PC2 scores for hESCs (blue dots) and hiPSCs (red dots) from Fig. 1. Each dot corresponds to one spectrum (pixel) of Raman mapping. Blue and red dots closely overlap, except for the top-left part of the graph which corresponds to positive PC3 and negative PC2 scores (b). Loading curves of (a) support that this PC3–PC2 region can be assigned to DNA/RNA bases, the typical frequencies of which are exhibited as positive bands in the PC3 curve (mainly 785, 1098, 1484, and 1575 cm^–1), and as a sharp negative band (785 cm^–1) for the PC2 curve. hESC human embryonic stem cell, hiPSC human induced pluripotent stem cell