Fig. 4

5hmC marks genes important for blood cell type function. a Using Euclidian cluster calling and the complete linkage method, 3208 genes (from cluster E in Additional file 2: Figure S4) were clustered according to their gene body 5hmC levels. The mean levels of gene body 5hmC from two biological replicates were used for granulocytes and CB-CD34+ cells. b Log2 of FPKM-averaged gene expression of all genes in the different clusters across the cell types. Only genes with FPKM >1 were included. Averaged log2 (FPKM) is shown ±95 % CI. *Two-tailed unpaired Student’s t test; p < 0.001. c Gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses of genes falling in the six clusters. Only the first two GO and KEGG terms are listed. Heat density plots showing gene-wise changes in gene expression (FPKM, y-axis) versus changes in gene body 5hmC (x-axis) between progenitor BM-CD34+ cells and mature blood cells: T cells (d), B cells (e) and monocytes (f). Only genes with FPKM and gene body 5hmC >1 were plotted. The logarithmic fold change in gene expression needed to be ≥2 to define increase or loss of gene expression in mature blood cells. In each plot, the Spearman rank correlation coefficient ρ and the number (n) of genes plotted are shown (exact permutation test for testing the null hypothesis of no correlation, two-tailed, p < 5 × 10⁻²⁰)