How to get unfiltered dge matrix which including empty droplets?

[echovv]

Hi Team,

Thanks for sharing such great tool!

I am trying to process a public dataset from GEO: GSE175421. I have successfully run through the pipeline and get the dge.txt.gz, which only include filtered barcode which are be though as true "cell". Actually my aim is want to get all barcode including empty droplet which is used to set as ambientRNA removal backgroud. Could you help me how can I achieve this point.
Second weird point is the cell number is far lower than my expect. Dge summary file mentioned there is only around 500 barcodeID. But I known there suppoused to have 2k-3k cell in each sample by checking the processed matrix uploaded by author. Do you think which step I may be wrong to cause it? Below I attached my running script.

script.txt

Thanks~
Ziwei

[alecw]

Hi @echovv ,

I can't tell what your pipeline is from the attached file. In a typical workflow, the aligned reads are processed by DigitalExpression to produce a DGE for each cell barcode that has at least 20 transcripts. That should be what you are looking for. Typically after the 20-transcript DGE is produced, another process filters the cell barcodes based on some determination of which are true cells. So if you are starting with reads, you should be able to use one of your intermediate files.