zero inflated correlation computation scRNAseq

Question

[sim.j.baum]

Hi all,
I have the following biological question:
Is target geneX coexpressed with genes in my gene list B in the scRNAseq data.
My approach would be to get the gene counts for each cell and then compute the correlation between geneX and my genes in list B.

What I was questioning myself - if the zero inflated property of the single cell RNAseq data would infer with the correlation values:

• For example give me a high correlation, because the majority of the cells are not expressing geneX and geneN from my list B, but cells which which express geneX do not express geneN - however the correlation value would be high because of the correlation of all non geneX, geneN expressing cells?

Other ideas to tackle the initial biological question in scRNAseq data are welcome,
thanks,

S

[GenoMax]

Have you seen: https://www.nature.com/articles/s41587-019-0379-5