FIG. 3.

sPLS-DA sample representations of the genomics (a) and proteomics (b) data. sPLS-DA was first trained on each data set [selection of 90 probe-sets in (a) and 21 proteins in (b)] on the 26 training samples (circles). For illustrative purposes, a second dimension was added but this second component is not relevant to discriminate the two classes. The testing samples (triangles) were then overlaid on the plot based on the prediction given by the sPLS-DA model. Late samples are shown with open symbols.