Fig. 1

Genomic properties of CTCF loops and flanking regions. a Profiles of H3K4me1 histone mark and gene and exon density across genomic regions within and around CTCF ChIA-PET loops in both MCF-7 and K562 cells. The loops and their flanking regions are split into bins each spanning 10 % of the loop length. For each bin, the median feature coverage for all loops is plotted. Profiles were normalized by subtracting the mean of all bins, displaying only the variation pattern across the profile. This was done because mean genomic bin coverage can vary substantially between chromatin marks and other genomic features, separating the profiles along the Y-axis and making pattern comparison more difficult. b Profiles of enhancer- and transcription-related chromatin states as defined by the ChromHMM algorithm, within and around CTCF ChIA-PET loops in K562 cells. Processed as above, but without normalizing the means as all the profiles have similar means. Weak transcription differs from the normal transcription state in that it is associated with transcript production but not with any further chromatin marks. c Expression level distribution for genes within and flanking CTCF ChIA-PET loops in K562 and MCF-7 cells. Flanking regions are equal in size to the loops they flank. Flanking genes located within neighboring loops are excluded from the set of flanking genes; there is no overlap between loop-enclosed and loop-flanking genes. Expression data are from the same cell lines as the corresponding loop sets. d Tissue Specificity Index (TSI, see methods) distribution for genes within and flanking CTCF ChIA-PET loops. Flanking genes are chosen as described above. e Coefficient of Variation (CV) distribution for genes within and flanking CTCF ChIA-PET loops. CV (standard deviation/mean) indicates the degree of variability in expression level of a gene across tissues. Flanking genes are chosen as described above