Large difference in mean reads per cell from cellranger and nCount_RNA from Seurat, is it a problem?

Question

[piloc16566]

I have single cell RNA-seq from PBMCs. Cellranger gives a mean reads per cell by dividing the total number of reads by the estimated number of cells. After loading into Seurat I can get the true mean reads per cell by taking the mean of nCount_RNA.

For the 5 samples I have, the mean reads from Cellranger are between 30k and 98k. The mean nCount_RNA is between 4100-4800.

Is this normal? Where are most of the sequencing reads going, is this from sequencing RNA from burst cells?

[antonioggsousa]

I never worked directly with CellRanger only with its final output. Usually what is imported to Seurat is the filtered result of the CellRanger pipeline. Are you sure that you're looking into the same files with CellRanger?

Did you check if the number of cells in each sample between Seurat and cellranger is the same?

[swbarnes2]

Are you sure you aren't mixing up reads and UMIs? Seurat is importing from a file of UMI counts.