Hello!

I want to analyze my ribosome profilig data coupling with RNA-seq (poly(A)) using DESeq2. So, ideally, my data contains only mRNA fragments but it's not true, cause for example Ribo-seq contains a lot of rRNAs, small RNA's that copurified with monosomes ad etc. In RNA-seq only 70-80% mapped on exons. I filter my data, but some trash rna's still remain.

So, I want to alnalyze DE and TE only among mRNA. Typically, when I analyse RNA-seq I start with that code:

#Create GFF object containing annotaition using GenomicFeatures:
my_gff <- makeTxDbFromGFF("Homo_sapiens.GRCh38.82.gtf", format="gtf" )
#Consider only exons:
exonsByGene <- exonsBy(my_gff, by="gene")

se <- summarizeOverlaps(exonsByGene, bamLst,...)

In case I want to consider only mRNAs in my DE and TE analysis, in what way I have to apply summarizeOverlaps (cdsBy or transcriptsBy or other variant) to count reads. Or maybe it is nessesary to use another package? Maybe it is better to use HT-Seq?

Thank you in advance,

Alex Gopanenko

You want to subset the genes to only include mRNA genes before counting?

Depending on what annotation you use, you could have different strategies. You could take exonsByGene and then subset to only those genes which have a CDS. Does that sound right? To match genes to CDS you could use:

tab <- select(txdb, keys=names(exonsByGene), columns="CDSID", keytype="GENEID")

Then by looking at unique(tab$GENEID), you would know which genes have a CDS. Because these are the same as the names of exonsByGene you can subset it by name.