Differential gene expression in triple-mutated AML. (A) Heat map of average normalized gene expression (Z-score) for each genetic group in the AML data set. Splits separate the different-gene clusters, and genes are sorted from high (top) to low (bottom) gene expression in triple-mutated AML. Only genes with an average expression of library-normalized raw-read counts higher than 30 are shown. Letter code as defined in Figure 1D. (B) Heat map of average Z-score normalized in GPR56 sorted fractions. Genes are sorted as in panel A. PP, GPR56⁺CD34⁺; MP, GPR56⁻CD34⁺; MM, GPR56⁻CD34⁻. (C) Enrichment analysis of biological processes for AML clusters. The background is defined using genes that are expressed and annotated to any ontology term. P values and odds ratios were calculated using Fisher's exact test, multiple-test correction using Benjamini and Hochberg method was applied to nominal P values. (D) Box and whisker plots of Z-score normalized gene expression for genes in clusters showing a gradual synergistic pattern compared with triple-mutated AML samples (ie, CL1-3 and CL8-10) for each of the 8 genetic groups. (E) Heat maps of the expression profile for transcription factors (rows) in the selected clusters shown in panel D in each genetic group (columns). (F) Normalized read counts for HLF, GLI2, and KLF12 in the defined, 8 genetic groups (left) and in the sorted fractions (right). Letter code defined in Figures 1E and 3B. (G) Combinatorial scatter plot showing gene expression (log[RPKM+0.001]) of HLF and GLI2 (left) and of HLF in combination with KLF12 (right) in AML and normal CD34⁺ populations. Symbols represent individual samples. (H) HLF mRNA expression (RPKM) in 8 matched diagnosis-relapse samples. Five pairs were triple-mutated at diagnosis and relapse (DNF), whereas 3 samples were NPM1 and DNMT3A double-mutated and gained an FLT3-ITD mutation at relapse (DN/DNF). Numbers indicate matched sample IDs.