How to merge paired-end reads from sam files?

Question

[aquaq]

Hi,

I have paired-end read sequencing data. I have aligned reverse and forward reads with bwa mem. Reverse and forward reads are 120 nucleotid long and they cover a 180 nucleotid long part of a genome, hence they overlap.

bwa mem  $REF $file1 $file2 -t 20 > $sam

When I open the sam output file, the first lines begin like this:

M00135:404:HBJFESJSN:2:1101:2016:1297   53      ref   ...
M00135:404:HBJFESJSN:2:1101:2016:1297   133     ref   ...
M00135:404:HBJFESJSN:2:1101:2646:1297   53      ref   ...
M00135:404:HBJFESJSN:2:1101:2646:1297   133     ref   ...

For every pair, I have the two lines aligned to the reference from the two directions ( I know, this is the normal output). Is it possible to combine reverse and forward reads to one sequence, thus getting a 180 nucleotid long alignment for each pair?

Many thanks!

EDIT: sorry for not being clear, I would like to merge pairs after alignment is done.

Answer 1

[WouterDeCoster]

BBMerge can do this :)

[aquaq]

Thanks. I have used pandaseq for this problem as well, but I would like to merge sequences after alignment, not before... I am sorry, I was not clear on this.

[WouterDeCoster]

I'm not sure what you biological motivation is for this objective, but I'm completely against tampering with alignment data. Which problem are you trying to solve?

[aquaq]

It would be just a trial. In a specific part of the sequence that we are interested in, there is a large number of mutations/sequencing error (it was a random sequence, but it was not supposed to be that random). I just wanted to be sure that it is not caused by some weird behaviour of pandaseq that I am not aware of before continuing with further analysis. But I could totally accept if that's unusual, I will find an other way to confirm it (eg by running bbmerge and comparing the results). Thanks for help!

[norah.saarman]

I would also like to do this, and yes, after alignment, because I am using a downstream application that needs a merged PE format, but the alignments contain < 1% of the total original fastq reads, and it will be much more efficient to merge only the aligned reads. Did you try using aftermerge? How did it go?

Answer 2

[WouterDeCoster]

I just saw this tool by chance, but obviously I have no idea how well it works.