Drug sensitivity prediction

We used the pRRophetic package in R to predict drug response based on pharmacogenomics and gene expression data using a ridge regression model. We obtained half-maximal inhibitory concentrations (IC50) for 138 different drugs in CGGA and TCGA patients, and then conducted Kruskal-Wallis and Wilcoxon rank-sum tests to compare differences between GBM patients with OBSCN or AHNAK2 mutations and those with wild -type genes. We also examined three different phenotypes based on OBSCN and AHNAK2 statuses (Double WT, Single WT, and Double Mut). To correct for multiple testing, we adjusted the p-values using the Benjamini-Hochberg (BH) method, and considered a false discovery rate (FDR) of less than 0.05 to be statistically significant.