Fig. 5: Unsupervised clustering of our mouse tumour transcriptional data identifies tumour clusters that relate to CMS biology.

a Barplots shows the conserved hallmark biologies across both species that are significantly enriched (padj < 0.05) in each CMS subtype versus other subtypes. b Elbow method on k = 1:10 indicated k = 3 (shown in vertical black dashed line) as optimal number of cluster (left). K-means clusters of n = 51 samples are visualised using principal component 1 (written as Dim1) and principal component 2 (written as Dim2) (right). c Heatmap depicts the predicted CMS class using each individual MmCMS classifiers with K-means clusters and genotype annotation on the top. d Schematic describing the development of the R-based MmCMS classifier.