Q: Callpeak and bdgcmp q value computation

[LinearParadox]

Use case
To my understanding, callpeak calculates q values across the whole genome, and considers each base an independent test. However, if using bdgcmp with -m qpois, only bases that are covered are considered. I just want to clarify that the behavior is indeed different between the two. Furthermore, would it not make sense to only consider covered bases as tests in callpeak instead of all bases within the genome?

[taoliu]

@LinearParadox For q-value calculation, we need to consider all the p-values from the whole dataset from independent tests for each base. And q-values are from the multiple test correction, or specifically the Benjamini-Hochberg process, on top of all p-values. And this behavior is the same as in bdgcmp and callpeak.