remove rRNA reads before DE analysis?

Question

[anubhav9]

Hi, I have a simple question.. I am doing RNA seq analysis on bacterial transcriptomes and most of my reads are mapping to rRNA genes (since the rRNA depletion is not very efficient). I am following a pipeline of FASTQC, aligning using bowtie 2, read counts with featureCounts and DE analysis with DESeq2.

I wanted to know if I should remove the rRNA gene rows from the count matrix before feeding it into DESeq2 or keep the count matrix as is ?

Thanks!

Answer 1

[LChart]

Leave it as is. You can exclude rRNA genes after the fact. The fact that most reads are mapping to rRNA genes suggests that you may especially want to keep rRNA genes, as without them the library size may be mis-estimated. However, you might care to evaluate the impact of including rRNA_fraction (e.g., rRNA_reads/total_reads) as a covariate in the DE model, as this may be a proxy for differences in sample collection or sample prep.

[anubhav9]

Thanks a lot!