Question

How to do FDR p.adj manually on "mr.edgeR object" from MEDIPS package?

0

Entering edit mode

pertille • 0

@pertille-9104

Last seen 5.8 years ago

Sweden

As I am working with a reduced-genome and the "p.adj" take in account all the windows in the genome, this filter is eliminating all of my windows with differential methylation profile:

Using this command in MEDIPS package, I can do my differencial methylation analysis including de FDR correction and consequently without significant windows results:

mr.edgeR = MEDIPS.meth(MSet1 = MeDIP_AV, MSet2 = MeDIP_C, CSet = CS, p.adj = "fdr", diff.method = "edgeR", MeDIP = F, CNV = F, minRowSum = 10)

On the other hand, if I do this command without FDR filter, I get a fair number of significant windows at the level of 0.005% (~300 windows):

mr.edgeR = MEDIPS.meth(MSet1 = MeDIP_AV, MSet2 = MeDIP_C, CSet = CS, p.adj = "none", diff.method = "edgeR", MeDIP = F, CNV = F, minRowSum = 10)

Then I eliminate the windows that have no methylation coverage in the mr.edgeR (p.adj="none") using this:

mr.edgeR <- mr.edgeR[mr.edgerR$MSets2.counts.mean != c("0"), ]

and now I wanna do the manual FDR correction in "mr.edgeR" object

But I don´t know how.

R Medips edgeR • 1.4k views

ADD COMMENT • link updated 8.8 years ago by Lukas Chavez ▴ 570 • written 8.8 years ago by pertille • 0

score 0 · Answer 1 · 2016-02-10

Dear Pertille, indeed, testing a huge amount of genome wide windows causes strong effects by subsequent multiple testing. To avoid many potentially unnecessary tests, MEDIPS allows to set an arbitrary integer to the minRowSum parameter of the MEDIPS.meth() function. I am aware that any value for this parameter will be somewhat arbitrary. Reasonable values depend on your library and reference genome size (please apologise that I cannot give the best recommendation for this parameter). However, MEDIPS will then only consider genomic windows tested for differential coverage when applying multiple testing. All the best, Lukas On 10 Feb 2016, at 18:42, pertille [bioc] <noreply@bioconductor.org<mailto:noreply@bioconductor.org>> wrote: Activity on a post you are following on support.bioconductor.org<https: support.bioconductor.org=""/> User pertille<https: support.bioconductor.org="" u="" 9104=""/> wrote Question: How to do FDR p.adj manually on "mr.edgeR object" from MEDIPS package?<https: support.bioconductor.org="" p="" 78139=""/>: As I am working with a reduced-genome and the "p.adj" take in account all the windows in the genome, this filter is eliminating all of my windows with differential methylation profile: Using this command in MEDIPS package, I can do my differencial methylation analysis including de FDR correction and consequently without significant windows results: mr.edgeR = MEDIPS.meth(MSet1 = MeDIP_AV, MSet2 = MeDIP_C, CSet = CS, p.adj = "fdr", diff.method = "edgeR", MeDIP = F, CNV = F, minRowSum = 10) On the other hand, if I do this command without FDR filter, I get a fair number of significant windows at the level of 0.005% (~300 windows): mr.edgeR = MEDIPS.meth(MSet1 = MeDIP_AV, MSet2 = MeDIP_C, CSet = CS, p.adj = "none", diff.method = "edgeR", MeDIP = F, CNV = F, minRowSum = 10) Then I eliminate the windows that have no methylation coverage in the mr.edgeR (p.adj="none") using this: mr.edgeR <- mr.edgeR[mr.edgerR$MSets2.counts.mean != c("0"), ] and now I wanna do the manual FDR correction in "mr.edgeR" But I don´t know how.<font face="sans-serif, Arial, Verdana, Trebuchet MS"> </font> ________________________________ Post tags: R, Medips, edgeR You may reply via email or visit How to do FDR p.adj manually on "mr.edgeR object" from MEDIPS package?