CNA score calculation per cell

Question

[Raghad]

Hello, I am analyzing single-cell RNA seq data, and I would like to get a CNA score per cell. I ran CNV estimation tools (like copykat and SCEVAN), and I get a CNA matrix output. I am interested in getting a CNA score per cell like they did here in figure 4A (3rd panel): https://www.cell.com/cancer-cell/pdf/S1535-6108(23)00174-5.pdf

Looking forward to your help.

Answer 1

[jared.andrews07]

Use infercnv, assuming you have reference "normal" cells and malignant cells of some variety in the same sample.

[Raghad]

I don't have normal cells in my tissue, all tumor unfortunately.

[jared.andrews07]

Not even a handful? Was the sample sorted in some way?

[Raghad]

Nope. I tried to look for normal cells using signatures, also by running CNA algorithms (like copykat and SCEVAN) but neither picked up any normal cells in my samples.

Yes they have been sorted for CD138+ (plasma cell markers), but maybe the tumor purity is very high..

[jared.andrews07]

Oh. What sort of cancer/cell type is it? I presume multiple myeloma? Do you get different populations/clusters? What technology are you using (10X, SmartSeq, etc)?

It's possible tumor purity is quite high, as I've seen very high malignant cell percentages in blood cancers when enriching for the malignant population (e.g. CD3+ cells in blood from Sezary Syndrome patients). Usually there are at least a handful of normal cells that get through regardless though. If you're using SmartSeq or such, where cells numbers are in the dozens/hundreds, then I could see it happening though.