Hierarchical Euclidean based clustering of the 2000 most variable loci. Samples split into 2 major cluster groups containing either CIMP+ve or CIMP-ve samples (major cluster 2 and 1 respectively). Major cluster 1 split into four sub-clusters. IDH1 mutated samples are highlighted with a black box. sGBM samples are labeled with their pair number (P#) followed by L to denote late lesion. pGBM samples are labeled with respective TCGA sample names.