Imputation when comparing multiome gex and atac?

[mitchTwoTimes]

Greetings,

I've struggled to find a definitive answer on whether I should impute the gene score when comparing to gene expression. For example, in the attached plot where the points are cells, the y-axis is expression for a single gene, and the x-axis is the matched accessibility, does it make sense to impute? FWIW the spearman's rho gets a little better with imputation.

Thank you

[rcorces]

I'm not sure I have a hard answer for this. I think its fine to impute though I could see this being disputed. Certainly it seems safest to not impute if it looks ok without imputation.