Entering edit mode
the consensus peak set extracted from diffbind is the counts of peak. if using the DESEQ method, diffbind will make analysis. so, can I directly use Deseq2 to analyze the consensus peak set. there are more other packages can analysis with Deseq2.
is there difference between the results made by Deseq2(with full count matrix ) and Diffbind(DBA_DESEQ2)?
The default mode in
DiffBindis to do aDESeq2analysis, so it is the "same" in that sense. However there are many ways to do an analysis withDESeq2, using different designs, contrasts, normalization, options to estimate dispersion, fit a model, perform difference tests, etc. So you can get one specific type ofDESeq2analysis out ofDiffBind, but using the reads counts to do aDESeq2analysis yourself allows you to have complete control over these choices.OK, I see. Thank you very much.
Hello, I am having a bit of trouble doing the contrasts with diffbind (I did post another question on this so hope this is nor cross posting), so it might be easier for me to export the count matrix made on the consensus peak to Deseq2. I have my count matrix like this :
K4_count <- dba.count(K4, bUseSummarizeOverlaps=TRUE, bRemoveDuplicates = TRUE, bParallel = TRUE). So if I understand correctly, I just need to makecounts = dba.peakset(K4, bRetrieve=TRUE)? And this matrix I can feed the DESEQ2 tool right? Sorry if I'm being obtuse, I'm not confortable yet with R or coding in any way (still learning).Thank you
Rita
This will work, but you should also set the
scoreparameter indba.count()toscore = DBA_SCORE_READSto get non-normalized read counts. The call todba.peakset()will return a GRanges object, with the counts stored in themcols()attribute.You can also get a fully-formed
DESeq2DESeqDataSetobject back usingdba.analyze()withbRetrieveAnalysis=TRUEafter doing a simple analysis, then use that to expand the design/contrast options directly inDESeq2.Thank you!