Big Difference cell estimation between emptydrops and cell ranger?

Question

[Sharon]

Hi All

I am working on sn/scRNASeq. I tried cell ranger v6 and v7, get cell estimates around ~11K, ~20k cells respectively, and some warning of law fraction of reads in cell.

However, when I used emptydrops I got around 60K, FDR was set to 0.01. I followed the tutorial below, but the difference is very big. Can anyone help explaining the difference, still new to sn/scRNASeq. Thanks

https://bioconductor.org/packages/release/bioc/vignettes/DropletUtils/inst/doc/DropletUtils.html