Could you clarify internal QC in CITE-seq-Count?

[marianaql]

Hi,

I was wondering if you could help me with the following urgent issue: our protein lane has 34k cells (after mapping with CITE-seq-Count using -cells option, no whitelist), RNA lane has 11k cells (after mapping with Cellranger using the filtered counts), intersect in barcodes is only 111 cells. We did see that the intersection between the raw Cell Ranger counts and CITE-seq-Count retrieved a very high (and nice) number. However, we know that we shouldn't use their raw counts, as these would be background (containing empty drops). In the context of CITE-seq-Count, I do not know whether using white-listed barcodes from mRNA is defensible in this instance - I do not currently understand whether the Cite-SEQ-count method fulfils some very important function that we should be paying attention to (i.e., cite-seq-count function is CRUCIAL to filtering for barcodes which are good quality). Is this akin to choosing to use the raw Cellranger output, containing cells which have already identified as background/emptydrops?

Thank you I really appreciate your time!!!

Mariana

[Hoohm]

Hello Mariana!

so the main issue with V3 is that the barcodes of the RNA and the Antibody barcodes don't overlap.

In other words, for each cell, you will have a cell barcode for the RNA and another one for the FB.

You can match them using the file found here: https://github.com/10XGenomics/cellranger/tree/master/lib/python/cellranger/barcodes/translation

There is a bit of code here that you can use to convert them.

https://gist.github.com/Hoohm/edf05b544105cd3f396d6e8c5dd66b22

Hope this helps

[Lopiniatre]

Hi @Hoohm, Thanks for the link toward this table. There are no header or description I could easily find, though I gather from your script that the RNA barcode is the first column and the corresponding antibody barcode the second column?