Hello, I am using the MEDIPS package to find differentially-methylated regions among three developmental stages, following the vignette, and have a question about how differential coverage us calculated. The differential coverage step, from the vignette: > mr.edgeR = MEDIPS.meth(MSet1 = DE_MeDIP, MSet2 = hESCs_MeDIP, + CSet = CS, ISet1 = DE_Input, ISet2 = hESCs_Input, p.adj = "bonferroni", + diff.method = "edgeR", prob.method = "poisson", MeDIP = T, + CNV = F, type = "rpkm", minRowSum = 1) Here, CSet is "coupling set", which is used for CpG-dependent normalization of the MEDIPS SETs. For my own run, I have designated a CSet, and have an Input comparator. What I am unsure about is whether, at this step, cytosine context is being taken into account. It doesn't look like it matters whether cytosines are in CG context or not, but since MEDIPS appears to have been designed specifically for human cancer studies, and my MeDIP samples are from plants, where cytosines in all contexts can be differentially methylated, I want to make sure that I'm not losing data at this step in the analysis. Thanks for help, --Kelly V.

Hello, In the given example, genome wide differential coverage between conditions is calculated based on the read counts (or mapping hits) per window; the returned counts, rpkm, p-values and fold changes are not influenced by any sequence context. On the other hand, the rms values are CpG density normalized and might be considered for comparison to bisulfite sequencing data or for quantifying methylation level. However, genome wide differential coverage between conditions is free of any CpG density normalization when the edgeR method is applied. Best regards, Lukas On Fri, May 16, 2014 at 9:17 AM, Vining, Kelly < kelly.vining@cgrb.oregonstate.edu> wrote: > Hello, > I am using the MEDIPS package to find differentially-methylated regions > among three developmental stages, following the vignette, and have a > question about how > differential coverage us calculated. > > The differential coverage step, from the vignette: > > mr.edgeR = MEDIPS.meth(MSet1 = DE_MeDIP, MSet2 = hESCs_MeDIP, > + CSet = CS, ISet1 = DE_Input, ISet2 = hESCs_Input, p.adj = "bonferroni", > + diff.method = "edgeR", prob.method = "poisson", MeDIP = T, > + CNV = F, type = "rpkm", minRowSum = 1) > > Here, CSet is "coupling set", which is used for CpG-dependent > normalization of the MEDIPS SETs. For my own run, I have designated a CSet, > and have an Input comparator. What I am unsure about is whether, at this > step, cytosine context is being taken into account. It doesn't look like it > matters whether cytosines are in CG context or not, but since MEDIPS > appears to have been designed specifically for human cancer studies, and my > MeDIP samples are from plants, where cytosines in all contexts can be > differentially methylated, I want to make sure that I'm not losing data at > this step in the analysis. > > Thanks for help, > --Kelly V. > > _______________________________________________ > Bioconductor mailing list > Bioconductor@r-project.org > https://stat.ethz.ch/mailman/listinfo/bioconductor > Search the archives: > http://news.gmane.org/gmane.science.biology.informatics.conductor > [[alternative HTML version deleted]]

Hi Kelly, the CSet is used for CpG dependent normalization ("rms" and "prob") only, and does not effect the differential coverage test results (logFC, pvalue ...) calculated by edgeR. To avoid confusion, you could set MeDIP to FALSE, then the CpG normalized values are not calculated. Best, Matthias On 05/16/14 18:17, Vining, Kelly wrote: > Hello, > I am using the MEDIPS package to find differentially-methylated regions among three developmental stages, following the vignette, and have a question about how > differential coverage us calculated. > > The differential coverage step, from the vignette: >> mr.edgeR = MEDIPS.meth(MSet1 = DE_MeDIP, MSet2 = hESCs_MeDIP, > + CSet = CS, ISet1 = DE_Input, ISet2 = hESCs_Input, p.adj = "bonferroni", > + diff.method = "edgeR", prob.method = "poisson", MeDIP = T, > + CNV = F, type = "rpkm", minRowSum = 1) > > Here, CSet is "coupling set", which is used for CpG-dependent normalization of the MEDIPS SETs. For my own run, I have designated a CSet, and have an Input comparator. What I am unsure about is whether, at this step, cytosine context is being taken into account. It doesn't look like it matters whether cytosines are in CG context or not, but since MEDIPS appears to have been designed specifically for human cancer studies, and my MeDIP samples are from plants, where cytosines in all contexts can be differentially methylated, I want to make sure that I'm not losing data at this step in the analysis. > > Thanks for help, > --Kelly V. > > _______________________________________________ > Bioconductor mailing list > Bioconductor at r-project.org > https://stat.ethz.ch/mailman/listinfo/bioconductor > Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor