Can I combine both single and paired end RNA-seq replicates for analysis?

Question

[bbyturner]

I have some older samples that were processed with single end RNA-seq and newer replicates that was sequenced using paired-end instead. Is it possible to process these two datasets and analyze them together?

Answer 1

[ATpoint]

It would be better to simply treat the paired-end samples as single-end by only using R1 for alignment/quantification, that will eliminate the single/paired bias. Do you have the fastq files available?

[bbyturner]

Thanks! Yep, I have the fastq files. I've been using Kallisto, would I basically just read in only R1?

[ATpoint]

Yes, that is what I'd do. Otherwise you create a mappability bias (paired-end may align "better" / differently than single-end), and as you cannot "upgrade" paired-end to single-end in silico you will need to stick to the "weakest link of the chain" which is "downgrading" the paired-end to single-end data by only using R1 as you say.