Hi everyone,
Let me introduce my data:
I am analyzing the single-cell RNA seq dataset. I'm gonna find the differential expression (DEGs) from different conditions in each cell type.
In addition, I'm working with Seurat's pipeline. My data is not suitable for pseudo-bulk DEGs analysis, therefore, mixed-model (MAST) is now my choice!
My question is, why do we need to use the normalized data (Scran-normalized, Log-normalized data) as the input of the MAST test, although the MAST test itself has the normalized method for count-depth (by cellular detection rate - CDR)?