Is It Acceptable To Merge Control Replicates When Doing Peak Detection In Faire-Seq Data?

Question

[Arturo_M]

Hello I am starting to work with Faire-seq data, I am using MACS to analize it. For the moment I only have have 2 replicates of the control experiment data, sequenced and aligned to the reference genome. The problem is that when I run MACS with one of the control replicates I do not find the peaks of some regions that I already know should be there...it is an excersice to "validate" my controls. Given this, is it valid to do a "merge" of both control data alignments and feed MACS with it, in order to get improved signal?

Thank you for your help

[Ying W]

What is your "control" experiment? Is it Reverse crosslink? Igg? a biological experiment of interest?

Answer 1

[Ian]

Hi, I have been interested in FAIRE-seq analysis, you might find this What Are The Best Practices For Faire-Seq Analysis? to a question i asked useful. The researcher I worked with was happy with the results of Fseq in the end. Only it is not trivial to produce the required mappability files.

I still remain to be convinced, but the opinion here is that a control sample is not required, especially using methods (e.g. Fseq, ZINBA) that use mappability. However, keeping to your question, if your controls are so different that they produce different results, it may indicate one is particularly different for some reason. I think merging biological replicates is usually a bad idea, but technical replicates is OK.

[Arturo_M]

Thank you, I tried to use Fseq but to produce the mappability files I understood I had to get a script from the Fseq developers, and I never got an answer from them. Do you know anything about that?