Fig. 5

Analysis of the placenta ChIP-seq ENCODE data. The 203 DMRs associated with pregnancy smoking status are depleted in H3K4me3 and enriched in H3K4me1 and H3K27ac, compared to 420 randomly selected regions^# on the Illumina array. The curves (a) show the mean profiles of ChIP-seq read counts in our 203 DMRs (red) and in the 420 random regions (blue) in triplicates. The heatmaps (b) correspond to the ChIP-seq data of one representative placenta sample for each mark. The box plots (c) show the distribution of areas under the corresponding curves. The values used for measuring the relative enrichments in histone acetylation and methylation marks of our DMRs are the mean of the mean enrichment over the region centered on the DMR ± 2 kb. A non-parametric comparison was performed using a Mann and Whitney sided U test. Heatmaps corresponding to all three samples are shown in supplemental files (Additional file 2: Fig. S3). ^#The “420 randomly selected regions” were obtained by running comb-p to select regions based on the null hypothesis “the number of identified regions is not associated with smoking status.” First, the DNA methylation levels were randomly redistributed, and then the analysis was performed using the same pipeline and parameters as with the real test groups, leading to the identification of CpGs and regions associated with smoking status by chance. In order to obtain a sufficient number of these random regions, the procedure was re-iterated 50 times