I'm jumping in here kind of late, but hopefully can help you out. The first
thing, like Brian suggested, is to make sure the inputs are reversed
(because the Charm data has unmethylated as the "enriched" sample), as
you're using MeDip data. However, since you are actually getting log-ratio
differences, I'm going to proceed like its correct.

The interpretation is not quite correct, because the methylation is
quantitative, not binary ("The region has 0.30427 times the risk of
being methylated
in samples of after treatment compared to samples of before
treatment."). It's more that samples after treatment have a logit
difference in methylation = 0.3. However, since differences in the logit
scale are not very interpretable, the easiest way to get % methylation
differences would be to take the anti-logit of your logit methylation
values, and then calculate the means on this scale, post hoc.

groups = ifelse(outcome==[whatever], 1, 0)
ilogit = function(x) 1/(1+exp(-x))
p = ilogit(methylation_matrix)
dm = rowMeans(p[,groups==1]) - rowMeans(p[,groups==0])

you can use the indexStart and indexEnd columns of the dmr list to find
each region's difference in mean on the %M scale.

Like Kevin suggested, you might want to prioritize some DMRs by this value,
as very significant logit differences e.g. -4 vs -3.5 is only a 1%
difference in methylation but a 0.5 difference in the logit scale.

Hope that help,
Andrew

	[[alternative HTML version deleted]]