Open sjfleming opened 3 years ago
Came here to post about this — I recently encountered this issue after running CellBender. ~900 / 100,000 called cells in my dataset have zero counts, the majority of which have less than 1000 UMIs. My stop-gap measure is to set a UMI cutoff at 1000 and simply continue the analysis, although I'll be keen to re-run CellBender once this bug is fixed.
Same here! I just noticed that my cellbender output filtered files contain several "cells" with zero counts
I see the same around 2% of my nucleus data has this issue this only seems to happen with low quality cells less then 1000 UMIs. So It's not a bad thing to actually remove those cells as they are poor quality anyway.
Occasionally we see a few droplets where the cell probability is high, but the posterior output counts are zero. Do not allow this to happen.
63