Use SAM file to pull reads from FASTQ

[pbm13]

Hi Folks,

I have a SAM file with unpaired reads (originally from a FASTQ) and I would like to use it to pull the read and its pair from the FASTQ file - does anyone know if there is a script out there to do this?

I have used the Picard tools SamToFastq but to my knowledge there is not a script in Picard or SamTools to do exactly what I described here (or maybe there is and I just haven't found it!).

Thank you!

[swbarnes2]

If the reads are unpaired, how can you pull their mate?

Juts work out the read names you desire, and write a short script to fish those reads out.

[sdriscoll]

^ good point...i have a feeling there is some miswording in the question.

I'd try to truncate the file through some form of filtering (maybe samtools or bamtools) and then use one of the sam/bam to fastq conversion scripts.

[pbm13]

Yes, sorry for the unclear wording. The SAM file is a result of mapping paired end reads to a reference. I have a SAM file with mapped mated pairs that I was able to convert to a FASTQ which worked great. But I also have a SAM file with mapped unmated pairs - it is with this file that I would like to use to pull the reads that mapped (but their "mate" did not) and their pair from the original FASTQ files.

Ideally the output would be these pairs in a FASTQ file.

[sdriscoll]

You should be able to extract those alignments as long as the aligner you used set the flags right. The unmapped mates will have 0x4 set and the mapped mates should have 0x8 set. You might need to name sort the bam first but then you could pull out only those reads with this:

samtools view -f 0xC -b alignments.bam > singletons.bam

Then convert that bam file into fastq.

[sdriscoll]

Actually that will also pull out all unaligned reads in addition to your singleton alignments. So more filtering will be necessary. Pairs make this tricky because the SAM annotation of pairs is messy.