Fig. 2.

Identification of Differentially Expressed Genes (DEGs) and molecular pathways between GBM subtypes. (A) Schema showing the number of samples obtained from publicly available databases (GSE119774 and GSE119834) or pd‐GBSC lines from the Human Glioblastoma Cell Culture (HGCC) resource. (B) Heatmap indicating the pairwise Pearson's correlations at the level of gene expression (RNA‐seq) between non‐tumoral cells, GBM cells or patient‐derived GBM stem cells (pd‐GBSCs) analysed in this work. Clustering of the samples was performed using the Ward.D method. Imputation of glioblastoma subtypes was obtained using the resulting RNA‐seq data in the GlioVis portal and the human GBM cell line LN229 was included for control purposes. Imputation of the MGMT promoter methylation status (U: unmethylated; M: methylated) was obtained using the R package mgmtstp27 (C) Violin plots indicating the coefficient of variation at the gene expression level calculated for the different primary bulk GBM subtypes. Asterisks denote statistical significance between cancer and control groups by means of a Wilcoxon Rank‐sum test (***: P‐value <0.001). (D) UpSetR plot representing the total number of differentially expressed genes (horizontal bars) and the potential overlaps (vertical bars) between non‐tumoral brain and the different GBM subtypes (adj. P‐value < 10⁻⁶). The number of samples and the number of differentially expressed genes (DEGs) identified in each of the comparisons are indicated. (E) Bubble plot illustrating the normalized enrichment scores of the different modules identified in the modular co‐expression analyses and the non‐tumoral tissue or Proneural (PN), Classical (CL) or Mesenchymal (MS) subtypes used in this work. Red and blue indicate, respectively, enrichment or underrepresentation of each of the modules in a given GBM subtype.