Hi everyone,
Nice to meet you all. I'm new to NGS data analysis. I have received H3K9me3 ChIP-seq. data, which peak calling was conducted by MACS2. I've observed that there are different enrichment pattern of the K9 signal (see below). I am concerning that the parameter for peak calling is not appropriate since the broad signal was called as separate small peaks. I know the --broad argument can be applied when handling broadly distributed histone modification signal, so that the nearby singal can be linked into a broad peak. However, if --broad is applied, will it affect the calling of those sharp peaks? Thanks.
: https://i.imgur.com/BhUQ8nZ.jpg Relatively sharp peak at promoter
https://i.imgur.com/yJ6VNFt.jpg Broad signal over gene body
: https://i.imgur.com/33eEeln.jpg Sharp peak at intergenic region
From my experience with ChIP data, it takes a lot of trial and error to find the correct parameters in order to identify your peaks. With the --broad parameter, I believe that there's an associate p-value cutoff parameter that's used when extending the peak from the initial peak region.
Looking at your ChIP signals, using the --broad parameter, you may end up extending peak regions too far as there appears to be a relatively large background signal. However, if your sharp peaks are well-resolved and have many reads mapping to them, it should work such that MACS2 identifies them perfectly even with the --broad parameter.
Also, consider running two types of analyses, if that works, i.e., one that identifies the narrow peaks and one that separately identifies the wider peaks.
Also, consider the other options such as HOMER, SICER, or bamCoverage.
Kevin