Hello,

Is there a way to test for differential binding above a log2 fold change threshold in the dba.analyze() portion of DiffBind? I understand we could use the fold argument in dba.report() to extract the subset of results that meet the desired threshold, but this is performed after differential testing. I am specifically looking for peaks that have >2-fold signal in one group over the other. According to Section 2.12 ("Differential expression above a fold-change threshold") of the edgeR user's guide, such hypotheses should undergo thresholded testing so that we can obtain meaningful p-values and FDRs.

I am using method=DBA_ALL_METHODS in dba.analyze() to obtain results from all routines. If thresholded testing is not possible in DiffBind, I could try to just export the relevant objects and perform the tests directly in the respective packages.

I am using DiffBind 1.14.6 and R 3.2.1, but I have also taken a look at the manual for the latest version of DiffBind.

Thanks for such a great resource and thanks in advance for your help.

Hello-

There isn't any way within DiffBind to call glmTreat(), which is the call you need to do what you want. As you suggest, you'll need to use edgeR directly to use this feature.

Regards-

Rory