Categorical phenotype

[CLVallerga]

Hi, is it possible to use coMethDMR to identify differentially methylated regions associated with dichotomous/categorical (e.g. case-control status) phenotypes?

[lxw391]

Thank you for your interest in using coMethDMR. To clarify, there are two main steps in coMethDMR: (1) identifying comethylatyed clusters (2) testing methylation levels in those comethylated clusters against a phenotype

For step (1), it can be completed for any type of phenotype. For example, if you have tumor vs. normal samples, you can compute co-methylated clusters using tumor samples which have more variabilities.

For step (2), coMethDMR currently only supports continuous phenotypes. However, one strategy for analyzing binary phenotype is to summarize each co-methyalted clusters from step (1) using median methylation level for example, and then use a linear model such as median (methylation level) ~ group status.

Hope this helps. Let us know if you have any additional issues.