Bulk RNAseq with limma-voom: estimateParam for controls vs disease

[dagkesamanskaya]

Dear Beate,

I am running powsimR for bulk RNA seq data from human brain samples (n = 150 (90/60 - disease/controls). First, I am running estimateParam function for controls and for disease separately.

As I get it from the documentation,

It is very important to supply the count matrix of a RNA-seq experiment profiling one group of samples or cells only. For example, for bulk please provide the count matrix containing the expression of the untreated samples / wild type ONLY;

Thus, later I run Setup function using control estimates as input in estParamRes and 'given' for LibSize parameter.

My control group is much more homogeneous than the disease samples, thus I believe the estimates from the control group will not reflect the distribution correctly. What do you think is wise to do? Is it possible to use the estimateParam results for both groups?