Fig. 4

Age-associated DNAm changes peak close to CTCF binding sites. a Pearson’s correlation of age with DNAm levels of CpGs within the amplicons of ELOVL2, PDE4C, and FHL2 are plotted for the blood samples of the training set (n = 38, blue) and validation set (n = 39, red). X-axis represents the position of CpGs within the amplicons. b Enrichment of CTCF binding at the position of these amplicons (gray shaded region) was then analyzed in chromatin immune precipitation (ChIP) sequencing data of hESC (GSM822297), K562 (GSM822311), and A549 cells (GSM822289). Peak heights were automatically trimmed by IGV tool (indicated in brackets). The positions of predicted CTCF binding motives are also presented. c Boxplot of normalized read counts of CTCF ChIP-seq data of A549, hESC, and K562 cell lines either at the 65 age-associated CpGs or at 1000 randomly chosen CpGs from 450 k BeadChip array. Read counts from ChIP-seq data were normalized by quantile normalization and analyzed within a window of 500 base pairs (p value was estimated by Mann-Whitney rank test)