I am aware that many studies use DMR to measure methylation level / region. However, DNA methylation impose its impact by suppressing gene expression. When we try to measure the magnitude / consequence of DNA methylation, should not we consider more about the correlation between methylation and mRNA expression (things like whether different CpG site have equal contribution on the expression of corresponding gene?) than merely defining statistically methylated region? I am a bit doubtful regarding the biological validity of DMR. Can anyone share some comment on this?
From cancer genetic perspective: Yes you are right. The possible consequence would be suppressing the target gene expression. You can do canonical correlation analysis on cancer data when you have both methylation and gene expression data for a set of samples, to see how target genes are affected by methylation in specific tumors. BUT defining DMR in cancers certainly has more advantages than impacting gene expression. DNA based markers like methylation are more stable than those of gene expression. Accordingly, there is a great interest in using these kinds of markers in cancer studies. See this fresh paper on how methylation data is useful!