Fig. 2: Multiple DNA Sequences Interact to Regulate Replication Timing.

a Hundreds of regions are controlled by multi-rtQTLs. b, c Two rtQTLs affecting the same region. Blue, yellow, and red lines represent one rtQTL. Purple and green lines represent the mean replication timing of individuals carrying the late- or early-replicating genotypes, respectively, at both rtQTLs. Considering both rtQTLs explains a larger fraction of variation (green lines are higher than blue lines; conversely for purple/red lines). Asterisks (in legends): any genotype at this rtQTL. In panel c, the GG/GG combination of alleles is associated with complete loss of initiation activity. d A replication initiation site associated with six rtQTLs. Each rtQTL was significant even after conditioning on all other five rtQTLs in the region. e, f rtQTLs exert additive effects. All regions with two (e) or three (f) rtQTLs were pooled; replication timing is linearly correlated to the number of early-replicating alleles (p values based on linear regression). e: n = 5416, 15,820, 23,551, 16,181, and 7158 genotype-replication timing pairs (for 0–4 early-replicating alleles, respectively). f: n = 770, 2136, 3844, 4310, 3520, 2039, and 661 genotype-replicating timing pairs (for 0–6 early-replicating alleles, respectively). For panels e and f, the top and bottom whiskers denote the maximum and minimum value, respectively. The top and bottom boxes denote the third and first quartile, respectively. The center denotes the median. Outliers were determined and removed using R boxplot option “outline=F”. g Multi-rtQTLs conform to a “promoter-enhancer” logic, primary rtQTLs being closer to the affected replication timing peak than secondary rtQTLs (two-sided Wilcoxon rank-sum test p-value indicated).