ribosomal genes in differentially expressed gene list

[liux]

a lab with partially degraded mRNAseq samples came back with a list of DE genes. And there are a few ribosomal protein genes in the list, which are not supposed to be there. How did this happen, and is there a way to 'fix' it, assuming the samples are not replaceable.

The analysis was done with RSEM after aligned to genome with STAR, and the DE analysis was done with EdgeR.

Thanks!

[liux]

any suggestion?

[blancha]

Since no one has tried to answer this question, I will try, to the best of my ability.

First, since you mention that the RNA was partially degraded, start by checking the RIN (RNA Integrity Number). If it's too low, just walk away from the analysis. Trust me. This is the best advice that any one can give on trying to make sense of degraded RNA.

Second, check how the ribosomal RNA was removed. Was poly-A enrichment used or a ribosomal depletion kit? One one occasion, I've had a researcher skip the ribosomal depletion kit. On two occasions, I've had researchers use the wrong ribosomal depletion kit or technique on Drosophila data.

Third, if the proportion of ribosomal RNA reads is significant, you have two choices to remove the ribosomal RNA reads in silico. You can either remove the rRNA reads before aligning to the reference genome by discarding reads that align to a FASTA file containing only the rRNA sequences. Or you have the option, with Cuffdiff at least, to discard the counts from the rRNA reads, after the alignment to the reference genome, with the option --mask-file.

If the proportion of reads from ribosomal genes is not significant, I wouldn't actually do anything, other than to remember to disregard the ribosomal genes in the downstream analysis.