Fig. 2

Bioinformatic analysis of RNAseq data from purified cell types. a Hierarchical clustering. The dendrogram of all RNA-Seq samples obtained with pvclust is shown with the bootstrap values on each node. Spearman’s correlation was used as a distance metric with the average linkage algorithm. b Principal component analysis (PCA). PCA plot of all RNA-Seq samples. Replicates of the same cell type are indicated by the same color as shown in the legend. c Maximum parsimony tree of all RNA-Seq samples with bootstrap values on each node. Scale bar corresponds to 500 hundred changes in binary gene expression. d Heat map of gene expression profiles shown with gene-wise (row) and sample-wise (column) clustering dendrograms. The colour corresponds to the square root of standardized TPM values (i.e., scaled by the standard deviation and centred to the mean) by row as shown in the legends on the left. Putative cell type specific gene clusters are shown on the right with the number of genes which belong to each cluster in parenthesis