Fig. 3

Clustering analysis using genome-wide DNA methylation and gene expression patterns in BCP-ALL patients with expressed fusion genes. a, b Principal component analysis of patients harboring non-canonical fusion genes involving ETV6 or RUNX1 together with patient samples with an established cytogenetic subtype (t(12;21)ETV6-RUNX1, t(9;22)BCR-ABL1, 11q23/MLL, and HeH) using a DNA methylation and b gene expression data. The fraction of the variance explained by principal component 1 (PC1) and principal component 2 (PC2) are shown on the horizontal and vertical axes, respectively. c, d Principal component analysis of BCP-ALL “other” patients using c DNA methylation and d gene expression data. Patients with DUX4-IGH rearrangements, ZNF384 rearrangements (ZNF384-r), and PAX5 rearrangements (PAX5-r), together with BCP-ALL “other” patients harboring other fusion genes or no detectable fusion gene, are shown in the figures. Patients that cluster together with IGH-DUX4- or ZNF384-rearranged patients, although with no characterizing fusion gene, are highlighted in the figure. The fraction of the variance explained by principal component 1 (PC1) and principal component 2 (PC2) are shown on the horizontal and vertical axes, respectively. e, f Unsupervised hierarchical clustering using e DNA methylation and f gene expression data. Patients belonging to the t(12;21)ETV6-RUNX1, t(9;22)BCR-ABL1, 11q23/MLL, and HeH subtypes are shown together with patients harboring t(12;21)-like fusion genes, DUX4-IGH, ZNF384, and PAX5 rearrangements. For ease of interpretation, only one of the paired reciprocal fusion genes detected in a sample is shown in panels e and f. The 1000 most variably methylated CpG sites or expressed genes were used in all panels