Fig. 1: Enhancers rely on distinct classes of TFs to create caQTL environments.
From: Context transcription factors establish cooperative environments and mediate enhancer communication
a, Schematic representation describing TF binding under different regulatory factor (RF) concentrations, with low concentrations as a result of independent binding and high concentrations of cooperative binding. Context-dependent differences in molecular readouts when SNPs impact TF binding are highlighted both at the level of individual ATAC–seq tags (middle) and enhancer-wide accessibility (acc.) changes at the population level (bottom). b, Process to prioritize SNPs based on their impact on enhancer-wide accessibility in cis. c, Comparison of absolute (abs.) SNP effect sizes against baseline enhancer accessibility in GM12878 reveals a bias of caQTL enhancers toward higher baseline accessibility. d, Schematic representation describing three ways of assessing TF function based on caQTL and DNA accessibility information. First, relating top BS strength to general peak accessibility (Methods; accessibility score, log-linear model, left). Second, comparing caQTLs to no-effect SNPs in terms of how frequently SNPs create a new best BS (Methods; initiator score, Fisher’s exact test, middle) and third, relating BS strength in the sequence context of SNPs to the SNP effect-size status while controlling for baseline peak accessibility (Methods; context score, log-linear model, right). e, TF context score (y axis, t-statistic) versus initiator score (x axis, log2 odds ratio). Each point represents a specific TF motif. Accessibility scores are overlayed with a blue–red color gradient. Colored boxes indicate the following three of the four TF classes: context-only (blue), context-initiator (green) and independent-initiator (yellow). High accessibility scores indicate USTs. Dotted lines represent thresholds used to define classes. f, TFBS specificity toward lymphoid-specific (LS) and tissue invariant (TI) enhancers using the DHS index. LS/TI ratios are compared across enhancers (n = 21,752) split by above- or below-average top BS strength (aBS and bBS) for a given TF. The y axis shows the log2-transformed ratio of ratios (LI/TI in aBS versus bBS; Mann–Whitney test, P = 2.2 × 10−15, P = 0.11, P = 1.1 × 10−14 and P = 1.2 × 10−5 for the comparisons of unlabeled motifs (n = 151, gray) to USTs (n = 120, red), independent-initiator (n = 18, yellow), context-initiator (n = 44, green) and context-only (n = 68, blue) motifs. Box plots denote the 25th percentile, median and 75th percentile. NS, not significant (P > 0.05). ****P < 0.0001.
