Statistical analysis
Statistical analysis of gene expression data was applied to log2-transformed values 22. A gene was considered differentially expressed between tumors and controls if the p-value was lower than 0.05 (FDR-corrected ANOVA). Sample clustering was performed using the k-means algorithm with 1000 random permutations and four different centroids. Centroid numbers were estimated using the elbow method and UPGMA hierarchical clustering; input data were scaled, centered and PCA transformed prior to clustering procedure. Differentially expressed genes were identified by analysis of variance (ANOVA) 23 between sample sets 1, 2 and 3. All analyses were performed using R (version 3.4.4), with the following packages: STRINGR, GGPLOT2, GRID, gridExtra, and GGBIPLOT.