Figure 3

Hierarchical clustering analyses of CRC test samples by expression levels of all splicing factors. (a) Unsupervised hierarchical clustering analysis of all 83 CRC samples based on the expression levels of all 280 splicing factor genes separates the samples into clusters with predominantly lower (blue boxes) and higher (red boxes) relative amounts of deviating exon usage than the average sample (black boxes), according to the oTIN subtype. (b) Samples considered to have the oTIN subtype were almost completely separated into two groups with low and high relative amounts of deviating exon usage after hierarchical clustering based on the expression levels of the total set of splicing factors. Both clusters were created using Euclidean distance metrics and complete linkage. MSS, microsatellite stability.