Figure 5

Supervised analysis using differentially expressed genes. In the three-dimensional principal component analysis (PCA) 99 samples are included. The signal used is PQN. The analysis is based on 1089 differentially expressed genes that were identified in a supervised way to distinguish between the 9 distinct leukemia subtypes. A sphere represents each sample's gene expression profile using the 1089-gene signature. The first three principal components (PC) account for 58.6% of variation of the data (PC1 = 40.3%, PC2 = 11.3%, PC3 = 7.01%). (A) Distinction by leukemia classification: spheres with the same colors represent the same leukemia subtype. (B) Distinction by sample preparation method: spheres with the same color represent samples processed with the same total RNA preparation method.