Entering edit mode
2.6 years ago
Josip
▴
40
Hi,
I'd like to compare two publicly available datasets that differ in the library prep method, and probably less importantly- read length. I am relatively inexperienced with RNAseq data analysis.
https://bmcbioinformatics.biomedcentral.com/articles/10.1186/s12859-017-1714-9#Sec8
This is the only piece of work I've found that tackles this issue. While I find filtering the reference transcriptome appealing, ratio-based correction seems quite deliberate.
Could the following workflow work?
Filtering the reference transcriptome-> alignment-> counting reads-> DESeq2 +/- sva
What is the aim of this comparison? Deciding what for your experiments is "better"?
No. Actually, I'd like to compare human bulk RNAseq of a FACS-sorted cell population from two different sites (and two different papers, libraries etc.) and learn what makes them 'different'.
I have preclinical, murine data that indicates these cells are significantly different at those two sites in regards to disease progression and treatment response. Clinical data is less definitive but is also pointing in this direction. So, I'd like to make the best of publically available, human datasets to further back up my preclinical observations.
That sounds like you're having a batch effect that completely confounds your two conditions (site 1 vs. site 2).
Unfortunately, in such cases, there's no way to determine what differences are due to biology vs. what differences are due to technical artifacts (like sequencing site).