I'm trying to figure out how to do a DMP analysis (using minfi dmpFinder) on a related sample (if it's even possible). Right now the code (not written by me) is:
dmps = dmpFinder(mVals, pheno = targets1$X8yrfactor, type = "continuous") pheno.exists <- !is.na(targets1$X8yrfactor) pheno <- targets1$X8yrfactor[pheno.exists] mVal.phenoNotNA <- mVals[, pheno.exists] dmps = dmpFinder(mVal.phenoNotNA, pheno = pheno, type = "continuous") dmps$probe = paste(rownames(dmps)) rownames(dmps) = NULL dmps = dmps[,c(6,1:5)] write.csv(dmps, file = "dmps_8yr_minfi.csv", row.names = F)
The sample size is 96, i.e. 48 twin pairs, which is why we want to be able to control for relatedness in the sample. By "twin pairs", I mean the methylation data we're using is from a sample of 48 pairs of identical human twins. At the moment I believe we're only going to be testing one at a time if we do find a way to test them on this sample.
Is there anyway to code it similarly to a mixed linear model where a randomization variable for each related individual is taken into account? I've been referred/found this, this and this resource but I'm not sure they capture what I need. Correct me if I'm wrong.
A simple linear model will not do this, which is why we're trying to figure out how to do this similarly to a mixed model. 'dmpFinder' identifies CpGs where methylation is associated with a continuous or categorical phenotype ('pheno', which we have multiple we want to test, we're trying to write the code first). 'targets1' is our cases (3 of them) that were removed following QC.