DGE on protein-coding and lncRNAs separately?

Question

[firestar]

I carried out differential expression analyses only on protein-coding genes because that is what I was interested in. Now after a while, I am considering looking into lncRNAs as well. Should I run DGE on just the lncRNAs or combine protein-coding and lncRNAs and rerun it together?

The number of genes analysed changes a lot in both these situations. This would also affect multiple testing correction.

[Satyajeet Khare]

Can you provide more details? You need to generate countmatrix of entire gene set and perform DGE analysis on that. You can shortlist genes of your interest (protein coding/non-coding) afterwards.

[firestar]

My choice of whether to run DGE on protein-coding genes/lncRNAs separately or together will change the total number of analysed features in my dataset. This will consequently change the stringency of multiple testing correction. Does this matter? My resulting DEG list based on p-values may be same in both cases, but a filtered list of DEGs based on q-values may be different.

Answer 1

[Nicolas Rosewick]

You should always use the most exhaustive annotation in DGE analysis e.g. ENSEMBL or Gencode (if available).

After DGE you can extract the lncRNA genes and look if some are differentally expressed.

[firestar]

Thanks for your reply. I suppose that would be reasonable if I was interested in all the annotations. Say if I was only interested in the protein-coding genes, would it still be better to use all the features?

[Nicolas Rosewick]

I think so. Which tool did you use for your analysis ?

[firestar]

DESeq2.......................