RNA-seq, TMM and rRNA counts. Should rRNA genes be removed prior to DE analysis?

Question

[gjuik]

When running a DE experiment (bulk RNA-seq), with each sample containing around 10% rRNA counts each, should the genes corresponding to rRNA be removed from the analysis. If they are not removed, will TMM trim them (considering that rRNA counts are greater than other genes' counts), and not include them in the calculation of normalization factors?

Answer 1

[Shred]

No, this will inflate the library size. If rRNA is not depleted using specific kit (RiboZero, for example), rRNA reads must be removed before performing any kind of normalization that uses the library size, like TMM, TPM and so on.

Source: TMM paper
DGE workshop