Figure 4

Accuracy of classifiers for independent platform with features ranked by the CV. 342 samples of exon-array and 76 samples of RNA-seq were used for each training and testing to predict the four GBM subtypes. The dotted brown line marks 90% accuracy.