Figure 3: CRC risk SNPs lie in highly recurrent VELs.
From: Hotspots of aberrant enhancer activity punctuate the colorectal cancer epigenome
(a) Co-localization of CRC risk SNPs and recurrent gained and lost VELs, non-recurrent VELs, and non-VEL enhancers. (b) Variant set enrichment (VSE) analysis showing the degree of enrichment of CRC risk SNPs (diamonds) in various genomic features relative to random SNP sets (boxplots). Red line indicates significance threshold (Bonferroni corrected P<0.05). (c) Normalized H3K27ac ChIP-seq tracks in normal crypt and CRC at the DUSP10 locus. SNPs are represented by vertical black lines, at middle, and linked to haplotype block structure, shown below. Red lines denote the genomic locations of recurrent gained VELs which colocalize with two distinct CRC risk loci containing the lead GWAS risk SNPs rs6691170 and rs6687758 (orange arrowheads). Each track is labelled with the sample type and name of CRC cell line, crypt, or primary tumor sample. N, normal; A, adenoma; B–D, Duke’s stage B–D; M, metastasis; U, unknown. Super-tracks in red correspond to median binned signal of all normal crypts or all CRC cell lines. (d) Maximum gained VEL recurrence at each recurrent gained VEL CRC risk locus. The lead SNP at each risk locus is shown on the x axis. Putative target genes are shown at the top of each bar. (e) Percentage of gained (red) and lost (blue) VELs that contain a risk SNP (lead or LD). G10+, G19+ and G30+ correspond to gained VELs recurrent in at least 10, 19 and 30 lines, respectively. L14+, L23+, L30+ correspond to lost VELs recurrent in at least 14, 23 or 30 lines, respectively. **χ2 P<1 × 10−5.
