Fig. 3
From: Gene clusters based on OLIG2 and CD276 could distinguish molecular profiling in glioblastoma
Gene clusters based on OLIG2 and CD276 generated by PCA analysis. 26 genes are obtained by random forest algorithm according to the expression subtypes (A). The full view of the locations of genes on chromatin is shown (B). PCA analysis revealed GBM subtypes can be identified clearly (C). Gene expression of three modules (module-classic, module-mesenchymal, and module-proneural) generated by PCA are shown in heatmap (D). Three gene modules expressed differently in four subtypes (E). ROC curve of the random forest algorithm for subtype classification is shown and the AUC reaches 0.855 (F)