Helllo, people!
I am using edgeR to analise my small RNAseq data. I got a miRNA GTF file from UCSC site and used htseq to count my reads. When I used edgeR for statistics, I realised the normalization is take acording to the GTF anotation, not from the all mapped reads. Of course there are reads that was mapped in different position and was not counted as miRNAs in my GTF. In this case the CPM normalization will be "reads/million of miRNA mapped reads" not "reads/ milion of total mapped reads".
I think my question is simple: should I use a GTF for all transcripts instead of a GTF file specific for miRNAs?
I think the best normalization is the one that consider the total number of total mapped reads because the number of miRNAs (or another specific molecule) could be variable in the samples.
What do you think about that?
Thank you
