Double Peaks For Both Forward And Reverse Tags In Macs2 Model Of Chip-Seq Data - Shall I Be Concerned?

Question

[Nick]

I analysed 3 chip-seq replicates (each with its own input control) using macs2. For two of the samples the model produced by macs2 looked fine. I am pasting an example below:

For one of the samples the model graph for both the forward and the reverse tags seems to exhibit double peaks:

This normally happens if you mix libraries of different fragment sizes but this is not the case. Shall I be concerned? What would you do?

[bede.portz]

Could you clarify a few things. When you say replicates, are you referring to the same cell type/tissue IP'd using the same antibody? If so, does one such technical/biological replicate give double peaks while another sample presumed to be identical give single peaks? Or, are the double peaks generated by some biologically relevant aspect of the experiment i.e. antibody, condition, etc? Also, lastly, are you seeing these double peaks in an input/control sample as well as an experimental replicate?

The point of these questions is to try to ascertain whether the double peaks are potentially caused by some biology, or by some technical issue.

Bede