We plan to perform RNAseq from whole transcriptome from FFPE tissue with 50 to 100 million paired-end reads mainly for DE analysis. The extracted RNA will be enriched by ribosomal RNA depletion and will yield degraded total RNA with RIN in the range of 2 to 3. Our question is: Which read length in the range from 50 to 300 bases you think is producing the best statistical reliable and biological meaningful results, considering that this kind of degraded RNA has an increased population of shorter fragments versus a lower population of longer fragments compared to non-degraded RNA. Does this kind of fragment distribution rather make results from shorter read lengths (e.g. 50b) more statistical valid, than performing longer reads (e.g. 150b) from very low fragment populations? Or doesn't this matter at all, since for shotgun sequencing the fragment lengths of degraded RNA with RIN 2 to 3 is still good enough to perform even longer reads?
To chime in again, please avoid acronyms or explain them: FFPE, RIN. This is an interdisciplinary community, and we cannot assume everyone knows about all molecular or computational acronyms.
Further, I think this post is at least borderline off-topic and should maybe better be asked on seqanswers.
Thanks for the link we will post there from now on. By the way this kind of RNAseq experiments produced very accurate results in the past, but in recent experiments more and more of our samples had to be excluded since their RNAseq failed the quality control. To minimize this sample loss we are looking to optimize parameters, including searching for the best read length for this kind of degraded RNA to improve their statistical validity.
In addition, I personally doubt that you will be able to base a statistically valid differential expression analysis on degraded transcripts, as you will mostly measure differential degradation. Ribosomal RNA depletion adds another noise source to the mix. Some sequencing facilities will also refuse to sequence that. As the approach might well fail, I would choose the cheapest option, the shortest read length should be ok as well, and run a small pilot first, not all the samples.
For such degraded RNA it's perhaps worth looking at a method like QuantSeq
Thank you very much for your help, I will discuss this with the sequencing facility, it looks as if this could be an improvement.