Fig. 4

A genetic contribution to variations in 3D chromatin conformation. a A CTCF Position Weight Matrix (PWM) is shown (Jaspar MA0139.1). Eight positions boxed by dashed lines have probability > 0.75 for a single base. We refer to SNPs at these positions as “motif disrupting SNPs.” Alleles matching the consensus base in the motif are labeled “strong motif alleles (S),” and alleles matching any other base are labeled “weak motif alleles (W).” b Boxplot shows the distribution of interaction frequencies at loops with exactly one anchor containing a CTCF motif disrupting SNP (N = 138), separated according to genotype. For each SNP, loop strengths are normalized to the mean value of the heterozygous genotype (WS). There is significant linear relationship between normalized loop strength and genotype by linear regression (p = 7.6e-5). c Aggregate contact map shows the average difference in interaction frequency per loop between SS and SW genotypes (top; N = 117 SNPs), and between SW and WW genotypes (bottom; N = 31 SNPs). The cross point of dotted lines indicates the 40-Kb bin containing the loop being evaluated. d Histogram shows the allelic imbalance in reads connecting loop anchors on the S vs W haplotypes in WS heterozygotes (N = 135 loops). The mean percentage of reads on the S haplotypes is significantly larger than 0.5 (p = 5.9e−4 by one-sided t test). e Line plots show signal of FIRE-QTL, INS-QTL, and DI-QTL by QTL genotype using 11 independent YRI individuals. Each plot shows the indicated phenotype as lines with light color, medium color, and dark color representing average signal across LCLs with the low signal genotype, medium signal genotype, and high signal genotype, respectively. For DI-QTLs, we split all 40-Kb QTL bins into two groups, based on the presence of either upstream DI bias (upper panel) or downstream DI bias (bottom panel). f For C-QTLs, an aggregate contact plot analogous to panel c is used to show the average difference in BNBC corrected interaction frequency (“Δ log(norm contacts)”) between the high and medium contact genotypes (top; N = 138 interactions), and between the genotypes medium and low genotypes (bottom; N = 94 interactions). The cross point of dotted lines indicates the 40-Kb test bin in question. g Boxplots show signal by QTL genotype using additional 6 individuals as a validation set. In each boxplot, three boxes with light shade, medium shade, and dark shade represent the average signal in the 40-Kb QTL bin from individuals with the low signal genotype, medium signal genotype, and high signal genotype, respectively. h Results of permutation test to evaluate the statistical significance of results in g. The solid vertical lines show the linear regression slope values obtained from the validation set (N = 6 individuals). The gray curves show the distributions of slope values obtained from 1000 random permutations. Corresponding bootstrap p values indicated in the upper left corner of each subpanel. i Line plot shows the fraction of “Y” QTL SNPs with nominal significance (p < 0.5) when tested for association with phenotype “X” (“nominal fraction”). Yellow diamonds show nominal fractions for phenotype “Y” QTLs. Red dashed lines show nominal fractions for all SNPs tested for association with phenotype “X,” and open triangles show nominal fractions for all SNPs tested for association with phenotype “Y.” Black circles and lines indicate the median and middle 95% range of 10,000 permutations in which SNPs were selected at random from the phenotype “Y” test SNPs. The number to the right of each line indicates the bootstrap p value (fraction of permutations with a nominal fraction higher than observed for “Y” QTLs). j QQ plot shows FIRE-QTL search results, including all SNPs tested for FIRE association (black points, N = 128,137), and several subsets of FIRE-QTL test SNPs as follows: SNPs also tested for DI association (light green, N = 46,784), SNPs also tested for INS association (light red; N = 6238), SNPs also tested for contact frequency association (light blue; N = 69,847), DI-QTLs (dark green, N = 152), INS-QTLs (dark red, N = 60), C-QTLs (dark blue, N = 53)