Fig. 1: Consensus clusters of gene expression data from Set 1.

Hierarchical clustering of normalized gene expression data revealed two distinct groups associated with BTK expression (A). The hierarchical clustering was repeated 10,000 times, taking a subset (80%) of samples and genes for every iteration to generate robust consensus groups. Consensus clustering was performed on the normalized expression data using 5000 genes with the highest median absolution deviation (B). The consensus clusters (k = 2) are significantly associated with BTK mRNA expression (C, BTK subtypes). There is significant difference in these BTK subtype’s probability of remaining untreated (progression-free) within this CLL cohort (D).