cell ranger count

Question

[archanaverma433]

Hi,

I have single cell RNA data and have one sample which I have resequenced. When I merged both samples from different runs, ran cell ranger count, I am getting more estimated number of cells in merged samples than sum of the cells from both runs separately , Is it right?

I have checked the number of reads of merged samples and they are equal of both runs reads.

1st run Estimated Number of Cells - 8,748
2nd run Estimated Number of Cells - 3,340 
merged run Estimated Number of Cells - 12,260

Thank you

[biofalconch]

Hi, could it be that maybe some cells pass the threshold of counts/genes one you have combined the data? (For example, if the count threshold is 150 for each cell and in both technical replicates a given cell has 149 in both, combining them will make this cell pass the threshold and be counted towards your estimated number)

[archanaverma433]

I will reframe the question.We have used same GEW well and did resequencing of the same library to increase read depth. That’s is why are worried why there is no overlapping and why we are getting more cells?

Answer 1

[ATpoint]

The knee method which is used to estimate the number of cells (that is the number of barcodes found frequently versus those found spuriously) is not a precise (in terms of absolutely accurate) procedure. After all, the knee method tries to find when the curve of frequency-ranked bacodes starts to drop, and that is really quite crude after all, so some minor differences are expected. So adding a resequenced sample is not going to precisely yield sample1 + sample 2 in terms of cells. It is very similar though, and I think this is normal and expected (and fine). Just continue.