Fig. 3
Correlation of the NCO recombination rate with epigenetic marks. a Heatmap and cumulative scores of histone and DNA marks. The impact of epigenetic marks on the NCO recombination rate was assessed by determining the average NCO rate (1 – D’2a,hap) for alleles located on tracks of histone marks (bivalent H3K27me3 & H3K4me1, H3K27me3, H3K4me1, H3K4me3, H3K4me2, H3K9ac, H3K27ac, H3K79me2, H3K36me3, H3K9me3), of 5-methylcytosine (5mC), and 5-hydroxymethylcytosine (5hmC), as well as of open chromatin defined by DNase I sensitivity. To avoid any bias due to the genomic location of the SNPs, the data set was divided into the respective genic sub-region (intergenic, upstream > 5 kb, 5′ UTR, intronic, exonic, 3′ UTR, downstream < 5 kb). Each combination of epigenetic mark and genic sub-region were tested independently using Mann-Whitney U test to determine if the overlap with the respective mark led to a significant difference in the NCO rate (Additional file 3: Table S2). Left panel: The color code in the heat map indicates the deviation from the average the NCO rate Δ(1 - D’2a,hap) of the respective sub-region in reference to regions without any mark (ranging from − 0.1 (blue) to 0.1 (red)). “ns” indicates a non-significant association; solid horizontal line separates recombination promoting marks (green arrow) from recombination repressing marks (red arrow). Right panel: The bars represent the average increase in the NCO rate score for each of the marks across the entire genome. ChIP-Seq tracks of histone marks were compiled from ENCODE data of various tissues and cell lines while tracks of 5mC, DNase I, and 5hmC marks were derived from the embryonic stem cell H1 [1, 21]. CO recombination hotspots marked by DCM1 tracks were excluded from the analysis. b Fold change in track overlap. The fold change in the overlap with 5hmC tracks (red), H3K27me3/H3K4me3-defined bivalent regions (orange), open chromatin-related DNase I tracks (blue) and 5mC-free tracks (green) and 5mC-containing tracks (gray) in response to the increase in the average NCO rate (1 – D’2a,hap) is shown. The dots indicate the fold change for the binned NCO rate average. Only singleton SNPs are shown. Dashed lines represent linear regressions; slope coefficient (m) and r2 value are indicated. Fold change for bivalency and DNase I was compiled from the ENCODE data set while 5mC and 5hmC were compiled from data provided for the H1 stem cell