Iterative Peak Overlap normalization and usage

[plbngl]

Hello!
I want to use the iterative approach to merge peaks across celltypes outside of ArchR, starting from macs2 calls and using the merge_peaks_iterative() function from BPCells (built on ArchR's logic).
My questions are about the ordering and nomalization of peaks:

1. Is it recommended to remove ovelapping peaks in each cell type and then do the removal in the merged set or is is better to do the removal all at once in the merged set?
2. How shall I normalize the -log10 Qvalues in each celltype? In the Science paper they mention a "score x million" normalization, but in the archR function it seems that there is a rank normalization (.getQuantiles)
   What is the recommended approach to use?
   Thank you !