Is it possible to compare 3SEQ and RNA-Seq data?

Question

[Kristin Muench]

I have two datasets corresponding to two conditions, each of which has been derived by a different sequencing method (RNA-Seq and 3SEQ). Each dataset contains multiple individuals, but no individual appears in both datasets.

Given that each of these sequencing methods introduces different biases (e.g., into the number of counts relative to the length of a gene), is there any way to do a differential expression analysis? Can one 'translate' RNA-Seq data into something with 3SEQ biases, or vice versa? Can one account for the biases statistically?

Thank you for any insights you can provide.

Answer 1

[Heather Vincent]

I think that the best thing to do will be to use only the counts mapped to the 3'-most exons in each dataset. I don't see how else you will be able to account for data that will be missing from the 3SEQ dataset

Answer 2

[Devon Ryan]

You have a batch effect that will confound things no matter what you do. If possible, you'd be best off to simply resequence things using the same technology (and do so at the same center at the same time if that's doable). If that's not possible then you're pretty much stuck with following Heather Vincent's answer. Aside from processing the RNAseq dataset differently, you might also look at normalizing to a large number of house keeping genes (perhaps try to create a covariate out of this to use in your model).