Im working with a scRNA data set and trying out some different filtering parameters other than using the Seurat vignettes. Specifically I am working at plotting metrics like count depth/cell and genes/cell. Im using this paper as a guide.

I plotted a histogram of counts per cell from my count matrix and got the these charts of the whole population and also zoomed in on the first 4000 counts/cells

My questions were:

1) What do the two peaks represent in this dataset/why are there two peaks

2) Based on the data from the zoomed in histogram, it looks like there's a drop at 900 before it starts to go up again. Should I filter out all counts/cell > 900 (or 1000)?

Thanks in advance!

Edit: Here is another plot of genes/cell from a different data set; why the low peak, then the high peak? What should be filtered from this set, should the filter be set at 300 or after the first small peak?