I'm doing a DNA methylation analysis using the limma package. The idea is to explore if there is a dose-response between exposure (pm2.5) and methylated CpG sites. Here is my code. I would like to know how can I do the linear assumption using the limma package.


dx <- with(phenoData.final, model.matrix(~ work_duration + Age + BMI + race))
csub <- combat.adjust[sample(seq(1, dim(combat.adjust)[1]), 10000),]
dc <- duplicateCorrelation(csub, design = dx, block = as.factor(Id.long))
pm.cont <- lmFit(combat.adjust, design = dx, block = as.factor(Id.long), correlation = dupcor.combat.fire$consensus)
pm.cont.eb <- eBayes(pm.cont)
pm.Cont <- topTable(pm.cont.eb,
                    coef = 2,
                    n = dim(M)[1],
                    sort = "p",
                    confint = 0.95,
                    adjust.method = "bonferroni")
  • $\begingroup$ Without knowing what your data looks like it's impossible to provide you with any advise. Do you have logit transformed data or are these methylation fractions? Why are you using limma rather than one of the many packages intended for finding differential methylation (granted, some of them end up using limma under the hood)? $\endgroup$ – Devon Ryan Dec 11 '19 at 21:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.