Fig. 3

Profiling diverse sequence tandem repeats and gene copy numbers in 46 colorectal cancers. a Clustering based on 225 microsatellites across four different classes. A 225 × 46 matrix including the presence (1) or absence (0) of microsatellite allele shift mutations was used for an unsupervised hierarchical clustering, which generated two clusters (MS Clusters 1 and 2). The heatmap of the two microsatellite classes (mono- and tetranucleotide repeats) with the most contributions are shown in two separate columns. b Clustering based on tumor/normal copy number ratio of 83 target genes. The median log2 ratios for all the target genes were used for an unsupervised hierarchical clustering, which generated two major clusters. Each major cluster has two subclusters. In the first column of the heat map, the MS Cluster identification for each CRC is indicated as a different color. The numbers on top of the heatmap indicate the chromosome where the genes are located. Copy number gain and loss are indicated with red and blue colors, respectively. c Log2 copy number ratio plots for all the CRCs having both MSI and CIN. For each CN index (x-axis), the log2 copy number ratio between read counts from tumor and normal samples (y-axis) is plotted. The median ratio value is indicated with lines of black, red, or sky blue, representing no change, copy number gain, or loss, respectively