Violation of Independence test for differential methylation (DM)

Question

[CY]

People say that the assumption of indenpendence is violated if using contengency table for differential methylation. How exactly does indenpendence is violated? Can anyone explain that a little bit? Thanks in advance

[German.M.Demidov]

Hi, it always depend how do you create the contingency table, but I think the idea is that neighbooring CpG positions have correlated methylation levels and simply summing them up (or testing them one by one and then applying a simple FDR correction) is not taking this dependency into account.

[CY]

Thanks for explaining. But when a single CpG site was tested, the contingency table sums up the (normalized) methylation counts of this CpG site from each sample (patient). I thought the independence we are looking for is the methylation level of this site between different samples, right?