I'm going through scATAC data analysis tutorial with Signac https://satijalab.org/signac/articles/mouse_brain_vignette.html that uses 10x data. The tutorial uses both the Feature-Barcode matrix (rows are genomic locations, columns are cells, and elements are counts) and the Fragments file(bed-like).
I was wondering why the analysis would need both of these data? I imagined that they convey the same information, since the Feature-Barcode matrix has the genomic locations. What am I missing?