Fig. 4

Differential expression analysis of Breast cancer and its subtypes. A Schematic overview of the samples used in our experimental set-up. B “Negative control” experiment comparing healthy test samples against the DGD (yellow) and controls for DESeq2 (gray). Twenty random test samples were chosen (shown as dots) and summarized by the boxplots and the mean (dashed gray line) for different cut-offs on adjusted p value (x-axis). As controls for DEseq2, we used the whole training set of breast tissue. We also tested with a random subset of 5 controls and show the mean as the solid gray line. C Enrichment score across breast cancer subtypes for driver genes and PAM50 genes derived from DGD and DESeq2. 20 random cancer samples were selected and compared to the DGD (yellow) and to DEseq2 using GTEx breast samples as controls (gray, see the “Methods” section for sample selection). The enrichment of cancer driver and PAM50 among the significant genes (adjusted p value < 0.01) is shown for each breast cancer subtype. D–G Breast cancer-specific differential expression analysis on a subset of ten marker genes, using 20 randomly selected samples. The box plots are colored based on whether the gene is known to be differentially expressed (red: upregulated; gray: normal expression; blue: downregulated) in a cancer subtype. The dots are colored based on the adjusted p value obtained by the DGD in each replicate