ChiP-Seq Biological Replicates for Broad Histone Marks

[Neberir]

Hi,

I have two replicates for five histone marks (three of them are broad chromatin marks). I would like to retrieve the overlapping peaks for the replicates for further analysis.

I tried calling peaks with macs2 for the replicates separately as well as by combining them. Since the peaks from the merged replicates would probably be skewed, I tried DiffBind (dba.overlap) with the separately called peaks. I was able to obtain the numbers of overlapped peaks for the replicates in a Venn diagram, but I failed to find how to write the overlapped peaks into a bed file.

Another suggestion I came across while trying to find another solution was IDR analysis, but I guess it is not recommended for broad chromatin.

Can someone tell me how to write the overlapping peaks from DiffBind to a file? Or any other suggestion would also be appreciated.

[rory]

I can offer some instructions on using DiffBind, but MACS2 may not be your best choice for broad histone marks. There are a number of other options out there (we use SICER). You may also want to look at the ENCODE project guidelines for settling on a set of broad peaks with two replicates.

To write a set of overlapping peaks to a file in DiffBind, the easiest way is to use the dba.peakset() function. If you have 10 samples, consisting of two replicates each of five histone marks, with the name of the histone mark in the FACTOR field of your DBA object, the following should work:

> consensus <- dba.peakset(DBA,consensus=DBA_FACTOR)
> mark1 <- dba.peakset(consensus,peaks=11,bRetrieve=TRUE,writeFile="mark1.bed")
> mark2 <- dba.peakset(consensus,peaks=12,bRetrieve=TRUE,writeFile="mark2.bed")
> mark3 <- dba.peakset(consensus,peaks=13,bRetrieve=TRUE,writeFile="mark3.bed")
> mark4 <- dba.peakset(consensus,peaks=14,bRetrieve=TRUE,writeFile="mark4.bed")
> mark5 <- dba.peakset(consensus,peaks=15,bRetrieve=TRUE,writeFile="mark5.bed")

[Neberir]

Thanks rory!