Search
Question: Diffbind: getting the raw reads
0
14 months ago by
veronique.storme0 wrote:

Dear Rory,

I found the solution to my previous questions. However, now I am comparing the results from the available edgeR analysis with the robust edgeR pipeline with QL method based on raw read counts. This is the pipeline that I want to use. I found the raw read counts by

reads.5 = dba.count(prol.5, peaks=NULL, score=DBA_SCORE_READS)
bindingMatrix.5 = dba.peakset(reads.5, bRetrieve=TRUE, DataType=DBA_DATA_FRAME)

However, these counts are different from:

de.5.edger = dba.analyze(contrast.5, method=DBA_EDGER, bFullLibrarySize=FALSE, bTagwise=TRUE, bReduceObjects=FALSE)
names(de.5.edger)
de.5.edger$contrasts de.5.edger$contrasts[[1]]$edgeR$counts[1:5,]

and as a result, I also get totally different DE genes.

How do I get the same counts with the dba.count function?

kind regards,

Veronique

written 14 months ago by veronique.storme0

Dear all,

dba.count(prol.5, peaks=NULL, score=DBA_SCORE_READS_MINUS)

I can see that when all negative values are put to 1, I obtain the same reads as in

de.5.edger$contrasts[[1]]$edgeR\$counts[1:5,]

Veronique

You can turn off the subtraction of control reads by setting bSubControl=FALSE in the call to dba.analyze().

-R