Hi everyone,

I have 3 groups in a RNASeq Experiment - A, B and C. Within each of these groups, I also have two genotypes, WT and KO.

I've run DESeq to determine differences among groups (without looking at genotypes yet).

Now, I know if I subset the DESeq object to contain only A and B groups, the sizefactors will remain the same as the full object (unless I re-run estimatesizefactors() again). But, what about the dispersions? Will the dispersions estimations of the subset object be the same as the full object?

The main idea of subsetting A and B is that group C is quiet different from these two and wanted to check if having C or not could somehow influence the dispersion estimates.

Thanks!

Dispersions will be estimated everytime you run DESeq(), existing ones already present in the DESeqDataSet will be replaced (the function even prints a message informing you about that) and will change when you subset. Does that answer the question?