Diffbind package DESeq2 analysis

Question

[ximena.garate]

Hi everyone,

I'm trying to run the Diffbind package with the results of an AgoIP sequencing experiment and when I run the differential expression I have zero differentially bound peaks. The script I used is just like in the manual example:

DBA <- dba.analyze(DBA).

Does anyone knows why I don't obtain any peaks with the DESeq analysis?

Thanks

Answer 1

[Rory Stark]

The most obvious answer would be that there may not be any significantly deferentially bound peaks in your experiment. This could be due to the biology (nothing is changing) or be related to the experimental design. For example, the experiment could be underpowered, mostly likely due to not having enough replicates to capture the variance in the data.

There are a couple of things you can look at. The starting point is to look at the MA plot:

dba.plotMA(DBA)

For more detail, you can get a report of all the peaks and their statistics. For example:

DBA.report <- dba.report(DBA, th=1, bCounts=TRUE)

Then you could look at DBA.report to see what the fold changes are and what the p-value and FDR rates are. You can also see the (normalized) read counts for all the replicates to see if there is very high variance. For example, the presence of a bad replicate could introduce such variance and lower confidence in detecting a change.

I am the author of DiffBind and often have a look at people's experiments when they get unexpected results. If you're interested you can post this issue to the Bioconductor support forum: https://support.bioconductor.org/

Please include which version of DiffBind you are working with.

Cheers-

Rory