Differential Gene Expression Analysis

Differential gene expression analyses for both microarray expression and RNA-seq data were performed using the limma (v3.42.2) package^{81 (link)}. For RNA-seq data, the read counts were first filtered to exclude nonexpressed genes, such that only genes were included for which at least three samples had a CPM (Counts Per Million) value above 1, i.e. genes for which sum(cpm>1)>=3. Secondly, read counts were normalized with respect to the trimmed mean of M-values (TMM^{82 (link)}) via the calcNormFactors function from the edgeR (v3.28.1) package^{83 (link)}, and then finally further processed using the voom function from the limma package. If otherwise not indicated, Box-plots for illustrating differential gene expression between groups of samples were generated in R using standard settings, such that the center line represents the median expression within the group, the box limits correspond to versions of the 1st and 3rd quartile, respectively, whiskers indicate the most extreme data points that are at most 1.5 times the interquartile region (IQR) above or below the box respectively, and points outside the whiskers are considered outliers.