Are the rlog or VST values from DESeq2 comparable across genes?

Question

[Xianjun]

Dear DESeq2 community,

I know I can use the rlog or VST (variance stabilized transformed) value from DESeq2 for sample clustering and visualization. I am wondering what else I can use it for. Can I use it as normalized expression value to compare expression between different genes, e.g. cluster genes into different groups? My guess is not, as it's not normalized by gene length. Please help to confirm. Thanks.

My another question is, from the DESeq result, I can get baseMean (which is the normalized count for ALL samples). My question is: can I get similar baseMean for each group of samples?

Thanks a lot,

-Xianjun