Supplementary Figure 4: Trac-looping PETs reveal high order of chromatin organization.

(a) Heat map generated by Juicer-box (Durant et al. 2016) showing the contact matrix (resolution = 500K bp) on whole chromosome 4 for Trac-looping and in situ Hi-C in resting CD4⁺ T cells. (b) WashU genome browser showing the read density of PETs with distance less than 150 bp (corresponding to accessible region), interaction matrices from Trac-looping data and from in situ Hi-C data, and distributions of compartment scores and boundary scores overlaid between pooled Trac-looping data (red lines) and the in situ Hi-C data (blue lines) for a 20M bp genomic region in chromosome 19 of resting CD4+ T cells. The interaction matrices, compartment scores and boundary scores were visualized at a resolution of 20K bp. Only informative PETs longer than 200K bp were used to infer compartment scores: 10M in total from Trac-loop data sets and 76M in total from in situ Hi-C, explaining the more fluctuation on the curve estimated for Trac-looping than for in situ Hi-C. (c) Smoothed scatter plot of the compartment scores calculated from Trac-looping data and in situ Hi-C data for all 20K genomic bins (n=154,783) of resting CD4 T cells. The r value is the Pearson’s correlation coefficient. (d) Smoothed scatter plot of the boundary scores calculated from Trac-looping data and in situ Hi-C data for all 20K genomic bins (n=154,783) of resting CD4 T cells. The boundary score is calculated between two neighboring windows (m and n) of 200K bp, sliding along the genome at a step of 20K bps, as the ratio between interacting PETs within m or within n to the interacting PETs between m and n. The r value is the Pearson’s correlation coefficient. (e) Percentages of CTCF motif orientations at CTCF anchored chromatin loops predicted by Trac-looping as a function of distance.