This is a test version of Biostars. For the public version, visit https://www.biostars.org.
DMRCate mean diff values

Hi, I am trying to get differentially methylated regions between cancer and normal using DMRcate, and my question is that I have a design matrix.

mod_our <- model.matrix(~as.factor(Status), data=meta)

This returns two columns where the first is the intercept (1 for all) and the second is as.factor(Status)normal which is 0 for cancer and 1 for normal samples.

Then I am running the following code:

Our_Data_DMRcate_M <- cpg.annotate("array", Our_Data_M_without_X, what="M" ,arraytype = "450K", analysis.type="differential", design=mod_our, coef=2)
Our_Data_DMRcate_M_dmrcate <- dmrcate(Our_Data_DMRcate_M, lambda=500, C=5)
Cancer_VS_NORMAL <- data.frame(extractRanges(Our_Data_DMRcate_M_dmrcate, genome = "hg19"))

My question is whether in this situation a positive mean diff value indicates more methylated in cancer or less methylated in cancer?

design methylation dmrcate
• 1,358 views • link
updated by james.hawley • james.hawley • written by Eren • Eren

1 answer

Based on your design matrix, where there is an intercept of 1 for all samples and a Status factor variable where 0 means cancer and 1 means normal, the mean diff value corresponds to the normal samples. So if the value is positive, the methylation will be higher in the normal samples than the cancer samples.

For a visualization of how this works, I'd recommend reading A guide to creating design matrices for gene expression experiments. Figure 1 has a good illustration of the "mean-reference model" you're working with.

You can (and probably should) double check this by plotting the methylation M-values at different loci, like positive control regions, negative control regions, and places where you see the largest changes in expression (up or down).

• 0 views • link
written by james.hawley • james.hawley

Log in to answer this question.

More posts like this