Compare peaks between clusters in sc-ATAC

Question

[Ahmed.waraky]

Hi

So If you have the called peaks per cluster from sc-ATAC and you want to compare between the annotated regions (promoters, enhancers, introns ..etc) between the clusters. Do you use raw peak counts? Do you run differential peak to background? Do you do percentage of annotated region per total cluster? As in each I'm getting different result. Also if I will do differential analysis, do I need to account for the GC bias?

Thanks in advance!