To identify the differentially methylated CpG sites, we analyze the Illumina 450K Methylation array output data using R-package minfi (Aryee et al., 2014). Raw image IDAT files were loaded into the R statistical computing environment using minfi, and the data was normalized using the SWAN function. This output a matrix of methylation indices (beta values), which were continuous values between 0 and 1 representing the ratio/fraction of the intensity of the methylated-probe signal to the total signal intensity for each probed CpG site. For each site, the methylation change was measured by the percent methylation difference (Δbeta) from paired baseline to DP/DD-DP samples. Minfi function dmpFinder was then applied to calculate the p-value of the logit transformed differential methylation. The p-values were corrected for multiple hypothesis testing with false discovery rates (FDR) q-values. We then used a cutoff of q-value ≤ 0.05 to define differential methylation
