Hello!
I am trying for the first time to run SICER for calling broad peaks in my ChIP-seq dataset. Many people suggested me to use this tool rather than MACS2 (with --broad option).
When running SICER, I am calling it with recognicer
command, since it seems to be the correct one for broad peak calling.
However, the default window size is only 200bp and I think it is a bit too short for broad peaks?
I am working on gH2Ax, which is a very broad histone mark. How would you suggest to adjust the window size?
So far, I decided to use 2kb as window size, which is the same value I was trying to use as bin size when running bamCompare/bamCoverage for generating .bigWig files.
Any opinion and suggestion is very appreciated!
Thanks in advance :)
First of all I suggest you use https://github.com/biocore-ntnu/epic2 which is a much more efficient reimplementation of the original SICER peak caller. I suggest you run with the defaults and then check on a genome browser.
I would not set the binsize for bamCoverage at 2kb, this is far too big. The tool will average all counts within that binsize so you cannot really see something. For a good browser track I would put it to 1bp. This gives the most meaningful representation. Be aware that bamCoverage only visualizes the sequencing data, it does not call any peaks. Feel free to post some screenshots of the browser tracks if you need advise.
Thanks a lot! I will then use epic2 instead.
You are right when you say "Be aware that bamCoverage only visualizes the sequencing data, it does not call any peaks." I am aware of it but I did not mention it in my post. I usually always use bamCompare normalizing the ChIP sample against its input. Unfortunately, in this dataset, the samples show a too large variation in the sequencing depth, which makes me result in large regions with negative values when computing the log2 ratio among the two.
Therefore I thought it was a better idea to proceed with bamCoverage instead.