Do paired end fastqs still have to be consistently ordered?

[Alexandre Amlie-Wolf]

Hi,

First off, thanks for developing an amazing tool! I'm processing some paired end data where the R1 and R2 files do not have the reads in the same order, and I found a post from 2016 where you wrote that STAR expects the paired fastq files to have the same orderings of reads (https://groups.google.com/g/rna-star/c/kwuP5XEfdPI/m/6msRLTQTBAAJ). Is this still the case as of v2.7.1a, the version I'm currently using? 

I haven't gotten any warnings about it or anything, and I'm currently running some tests comparing sorted vs unsorted input, but figured I would check from the source!

Thanks,

Alex

[Alexander Dobin]

Hi Alexandre,

yes, absolutely, the Read1 and Read2 files have to be ordered perfectly.

STAR does not check the ordering and if it's not maintained, it will not map improperly paired mates.

[Alexandre Amlie-Wolf]

Hi Alex,

Makes sense and is definitely consistent with the tests I've been running. Thanks!