I am using DiffBind and trying to export the matrix of values for all peaks. If I use dba.report() with bcounts=TRUE, I get those values, but it's only for the two groups being compared. What if you have more than two groups?

I tried dba.peakset() with bRetrieve=TRUE. That seems to return what I want, but the actual values are different than the ones from dba.report(). I assume they are just normalized differently, but how can I normalize them the same way?

You can change the read score values using the score parameter in dba.count(). If you set peaks=NULL, the score will be recalculated without requiring the read to be re-counted. See the man page for dba.count() for all the possible score values.

Here are some examples:

# Raw read counts
> mDBA <- dba.count(myDBA, peaks=NULL, score=DBA_SCORE_READS)
> allReads <- dba.peakset(myDBA, bRetrieve=TRUE)

# RPKM values
> mDBA <- dba.count(myDBA, peaks=NULL, score=DBA_SCORE_RPKM)
> allRPKM <- dba.peakset(myDBA, bRetrieve=TRUE)

# TMM-normalized read count (via edgeR), 
# using read-in-peaks adjusted to be counts-per-million
> mDBA <- dba.count(myDBA, peaks=NULL, 
                    score=DBA_SCORE_TMM_READS_EFFECTIVE_CPM)
> allRPKM <- dba.peakset(myDBA, bRetrieve=TRUE)

You may not be able to get the exact score values returned by dba.report() for a specific contrast, as those scores are normalized based on the specific analysis method (eg right now you can't get DESeq2 normalized score values using dba.count()) as well as the specific subset of samples involved in the contrast.

-Rory