Fig. 3

Disruption of certain motifs is associated with cis effects. a Percentage of shared motifs in tiles that show cis effects vs. those that do not. p value shown is from a one-sided Mann-Whitney test. b Plot showing the activating motifs whose disruption is significantly associated with cis effects (FDR < 0.05). Left: effect size associated with motif disruption. Middle: additional variance in MPRA activity explained by the TF. Right: enrichment of a given TF motif across biotypes, as determined by a hypergeometric test. Black dots denote significant enrichment (FDR < 0.05). The ETV5 TF has two “best” motifs according to the curated Lambert et al. [30]. TF list, and therefore, the average of these two motifs are plotted, with the bootstrapped 95% confidence interval shown. c Relationship between cis effect sizes and the ETV1 motif, where “maintained” are sequence pairs that both have the ETV1 motif, “disrupted in human” are pairs where the ETV1 motif is present in mouse but not in human, and “disrupted in mouse” are pairs where the ETV1 is present in human but not in mouse. A cis effect size > 0 indicates the mouse sequence has higher activity whereas a cis effect size < 0 indicates the human sequence has higher activity. p values shown are from a two-sided Mann-Whitney test. d Genome browser screenshot of an example locus showing a cis effect. Only motifs that were found to explain ≥ 1% of the variance in MPRA activity are shown. e MPRA activities for human sequence (orange) and mouse sequence (green) in hESCs and mESCs for the locus shown in d. p values shown are the q values calculated by MPRAnalyze