Understanding peak calling on whole dataset (piror to cluster identification)

[newtonharry]

Hi,

So I'm trying to call peaks on the entire dataset and I've gone ahead and also performed the merge_peaks function as well. I've had to adjust it to suit the data it expects (a dictionary with the values being peak dataframes for each cluster). I've found through experience and in a related post that the peaks end up being around 501bp each in my case, every time. Since I'm not using these peaks for analysis, but in assisting to define the feature set of the data, is the peak merging step still important here? Should I merge it a different way considering the variable peak width is important to me.

Thanks!