Question about proportion of genes assigned to grey module

[ddressman91]

From my exposure to classic WGCNA and this package so far, it seems like classic WGCNA often assigns most genes to a module and leaves a very small grey module (especially when using filtered genes as input). The vignette for this package, and my runs with my own data, tend to have the majority of input genes in the grey module (even with input filtering, and even when running this package using the pseudobulk vignette). Do you think the nature of single-cell data as a sparse matrix is the main reason for the differential size of the grey module in classic WGCNA vs. this package? Or are there other possible reasons for this phenomenon?

[smorabit]

This is a good question, and this is an obvious difference when performing WGCNA on bulk RNA-seq vs. single-cell. We have not done extensive testing to determine exactly why this is the case, but in general, the grey module contains genes that have a lower variance or lower expression level compared to genes in other modules. I think it could be the nature of single-cell data being sparse as you mentioned.