Hi,

I am working with snRNA-seq data and performing pseudo-bulk differential gene expression analysis using DESeq2.

To do this, I did pseudo-bulk (aggregate counts across the cells). Then, I selected a specific cell type (e.g., Neurons). Then, I used DESeq2 to compare differential expressions between cases and controls within specific cell type (e.g, Neurons).

However, I have noticed that my cases have fewer neurons compared to controls. My question is:

Does performing pseudo-bulk and differential gene expression of specific cell types account for differences in cell type abundance across samples (i.e., fewer neurons in cases vs. controls)?

If DESeq2 does not correct for this, what are the best approaches to adjust for cell type proportion differences in my differential gene expression analysis?

I greatly appreciate any guidance on the best normalization or modeling strategies!

Thank you.

The short answer is yes. When you compute pseudo-bulk estimates, the counts/gene are expected to be somewhat proportional to the number of cells you have aggregated, which in turn means that the counts/sample for the pseudo-bulk data should also be proportional to the number of cells. When you fit the model using DESeq2, an offset is incorporated in the model to account for library depth, which we assume is proportional to the number of cells aggregated, and therefore should adjust for differences in the number of cells.