The value of 'Fraction of high-quality fragments overlapping peaks' is greater than 1

[haloudashu]

Hi Kai Zhang:
I am using snapatac2 version 2.7.0.
After running snap.tl.macs3(), I can view the peaks in adata.uns['macs3_pseudobulk']. I noticed that there are some rows in the peaks with identical "chrom", "start", and "end", although their other columns like "score" differ. Since peaks should be unique, I deduplicated the peaks in adata.uns['macs3_pseudobulk'] to keep only rows with unique "chrom:start-end" combinations for subsequent calculations.

Running snapatac2.metrics.frip(adata, {"n_frag_overlap_peak": peaks}, normalized=False) gives me adata.obs['n_frag_overlap_peak']. I then divided adata.obs['n_frag_overlap_peak'] by adata.obs['n_fragment'] to calculate the "Fraction of high-quality fragments overlapping peaks" for each barcode. I found that 2709 barcodes have a "Fraction of high-quality fragments overlapping peaks" greater than 1, which is clearly unreasonable. Although these barcodes are not identified as cells, the fragment counts for each barcode in adata.obs['n_fragment'] may be inaccurate. What could be the reason for this?

Thank you!