Could the UCell module score be used as a metric for co-expression within your gene set?

[Ruth-hals]

Hi, thank you for developing this tool, very useful!

I'm interested in comparing coexpression levels between different gene sets. Could I use the module score as a metric for co-expression within a gene set? and then compare these amongst different gene sets? I could then use this as a read-out that genes in gene set A are more co-expressed than in gene set B based on pairwise test for all module scores?

Thanks for the info, Cheers, Ruth

[mass-a]

Hello Ruth,

I am not sure one can interpret UCell scores as a measure of co-expression. For example, you may have a simple signature made of genes A and B, and two cases; 1) gene A is highly expressed and gene B has zero expression; 2) gene A and B have both medium expression. You will in generally obtain similar UCell scores for case 1) and 2), therefore they are not a direct evidence of co-expression.

You may want to look into methods specifically developed for single-cell co-expression analysis, such as WGCNA or others.

Best -m