Dear Dr. Patro, developers of Salmon and BioC community,
I have put great amount of effort to find out answer to this question on internet. Couldn't find it anywhere (bioconductor, biostars, seqanswers, github etc.).
I have 3 replicates for each sample and one of the replicates from each sample is single-end and other two are paired-end as SE and PE were processed at different facilities (I know I have to do batch correction in downstream analysis). Now I want to use transcript level abundance from quant.sf file which I derived for each replicates using Salmon's quasi-mapping pipeline (used appropriate flags for SE and PE reads). All these (SE & PE) reads are strand specific.
My question is, can I use quant.sf directly from these replicates for downstream DE analysis using tximport or do SE or PE requires separate kind of processing before I can use them together as replicates for downstream analysis. I am planning to use limma-voom for my DE analysis.
Thank you so much for your time and apologies if the question was answered already.
Sandip Darji
Thank you so much Dr. Love for your prompt response.
I am including "batch" vector (batch 1 for SE reads and batch 2 for PE reads). So to confirm, you can treat quants from biological replicates as if they are normal replicates and it doesn't matter whether they are SE or PE. In other words, you can have biological replicates from SE and PE and treat them together as any other normal replicates. correct? I knew the fact, while dealing with raw counts (e.g. featurecounts or the counts from star-aligner), one can treat both SE and PE the same way as if they are all from same library (all from SE or all from PE) as long as they are biological replicates. Just wasn't sure the same is true while dealing with quants from salmon.
Thanks again,
Sandip Darji
Yes, that's what I meant
Thanks, again.