Must paired-end reads be in the same order in the two files for Tophat?

[friducha]

Hello,

Will TopHat work if R1.fastq and R2.fastq have their reads in different order?

What if R1.fastq has some reads whose R2 mate did not make it past QC, and viceversa (R2.fastq has some reads whose R1 mate did not make it past QC)?

I could only find this information about how it uses paired-end information after mapping independently each mate, but I can't find info on how it relates mate pairs to each other on the two files.

TopHat maps left and right reads separately using Bowtie, that is, it doesn't use Bowtie's pair searching like --fr, --rf, --ff. Using the mapped reads, TopHat finds pairs if the two reads of a pair are on different strand (it ignores if they are on the same strand) and the inner distance is within user specified range.

From: http://seqanswers.com/forums/showpos...49&postcount=2

Thanks for your input.

[Brian Bushnell]

Reads always need to be in the same order. It is best to use tools like BBDuk that keep pairs together for quality control; doing QC on the two files independently will only cause problems. I did write a tool for fixing that situation, though, repair.sh.

http://seqanswers.com/forums/showpost.php?p=141460&postcount=45

[SES]

Just for reference, there are quite a few tools that respect paired data, such as Prinseq and Trimmomatic. Often, custom pipelines don't take this into account. Also, Pairfq is a lighter approach to pairing reads. This makes it really easy to incorporate into a pipeline. In most cases, the follow is all you need to install:

curl -L git.io/pairfq_lite > pairfq_lite
chmod +x pairfq_lite
./pairfq_lite -h

The last command just prints the usage, which is explained on the the wiki or from the inline documentation available at the command line.