Fig. 5

ZNF263 binding is overrepresented in hypomethylated regions both in silico and in vitro. a Shows the most similar sequences within the co-methylated calcium channel DMRs. b Bars show the number of hypo- and hypermethylated DMRs among the p < 0.0001 subset (n = 852) having significantly similar sequences to the calcium channel consensus sequence presented in a. Q-value thresholds represent false discovery rate-adjusted p values attained from log-odds scores of the sequence similarities. A chi-square test was used to compare the observed distributions at q < 0.05 and q < 0.1 to the expected distribution, as shown by the percent hypomethylation of the original DMRs. c The calcium channel consensus sequence was significantly similar to three predicted zinc-finger transcription factor binding site sequences (motifs). d Shows the percent representation of hypomethylated DMRs that overlapped with transcription factor binding sites as measured by chromatin immunoprecipitation sequencing and posted within the Encode project. To allow for a progressive increase in genomic window size, the top 300 hypomethylated DMRs were compared to the top 300 hypermethylated DMRs, where the dotted line represents the expected overlaps under random conditions. Note that the ZNF263 binding occurs close to the hypomethylated DMRs, while the EGR1 binding is more distantly located, becoming significantly associated only at a 5 kb distance. ****p < 0.0001 and ***p < 0.001 represents chi-square tests. The consensus/motif analyses were performed using the MEME suite tools; a = MEME, b = TOMTOM, and c FIMO (http://meme-suite.org/)