DMRcate

DMRcate is an innovative R software package designed to identify and characterize differentially methylated regions (DMRs) across phenotypes in the human genome, focusing on data from the Illumina HM450K BeadChip (450K array). The method employs a Gaussian kernel to spatially smooth replicated methylation measurements across the genome, efficiently identifying and ranking DMRs without being influenced by genomic annotation or the direction of differential methylation signal changes. This approach overcomes the limitations posed by the irregular spacing of methylation sites and provides statistical significance to each identified DMR by comparing it against a null model. Comparative analyses with both simulated and real data demonstrate DMRcate's superior predictive performance over existing methods like Bumphunter and Probe Lasso and show its efficacy comparable to comb-p. Validation of array-derived DMRs against DMRs identified from whole-genome bisulfite sequencing of the same DNA samples further underscores its reliability and accuracy. DMRcate represents a significant advancement in epigenetic research, offering a balanced approach to DMR identification that combines differential methylation signal smoothing with precise threshold specification. Although the 450K array's design preferentially targets CpG sites likely to be differentially methylated, DMRcate's findings highlight the array's limitations in capturing the full depth and complexity of methylation signatures compared to sequencing technologies.