There is a large discrepancy between the results obtained from pseudobulk methods and mixed models. #145
Description
With pseudobulk methods, very few genes reached an FDR < 0.05 (all cell types; in total 7 genes), whereas mixed models yielded many significant results (2900 genes).
My sample design is 2 vs 2, with two samples in the experimental group and two in the control group. My goal is to investigate how the experimental treatment affects gene expression.
Is this normal? Which results should I use?
Similarly, regarding stagewise analysis, should I rely on pseudobulk methods or mixed models? Some results that were not significant with pseudobulk methods became significant after stagewise analysis.