champ.DMP uses the model ~ 0 + pheno (line 20) while champ.DMR(method=="DMRcate") uses ~ pheno (line 325). I believe the second is required as the DMR method uses cpg.annotate from the DMRcate package which itself requires a model with an intercept (stopifnot(colnames(design)[1] == "(Intercept)"))
Thus, the list of significant probes that goes into DMR is not the same list of probes identified in DMP. It would be ideal if these two methods allowed you to specify your model or alternatively, if they both used ~ 0 + pheno as this is a more common design.
champ.DMP uses the model
~ 0 + pheno(line 20) while champ.DMR(method=="DMRcate") uses~ pheno(line 325). I believe the second is required as the DMR method uses cpg.annotate from the DMRcate package which itself requires a model with an intercept (stopifnot(colnames(design)[1] == "(Intercept)"))Thus, the list of significant probes that goes into DMR is not the same list of probes identified in DMP. It would be ideal if these two methods allowed you to specify your model or alternatively, if they both used
~ 0 + phenoas this is a more common design.I have also posted a question on Bioconductor to alert DMRcate to this issue. https://support.bioconductor.org/p/123226/