Hello,

I would like to estimate the differential usage of non-coding portions of the transcripts in RNA-seq experiments between two conditions that were derived with non-polyA depletion of rRNAs (e.g. RiboZero), which therefore maintain introinc reads. Moreover, I am interested in the differential usage of the UTRs (5' and 3'). 

My question is: should I include the counts from different portions of the transcripts (coding+non-coding) and analyze the whole information aggregated or should I analyze separately coding and non coding regions (which have different, intrinsic, characteristics, such as increased feature length compared to exons)?

Thanks,

Stefano