how to extract both ends mapped to different reference from bam

[jmartin]

Is there an easy & robust way to extract from a mapping bam file all the read pairs where both ends mapped, but to different reference pieces?

I had thought it might be possible using:

samtools view -F (or -f) <some flag>

but I can't figure out exactly what flag to use. Is there a bit for read pairs mapping to different reference pieces?

[dpryan]

The simplest method would be with awk, since there's no flag to designate this:

samtools view foo.bam | awk '{if($7!="=") print $0}' > different_chromosomes.sam

Note that the produced SAM file is missing its header, but you can change the samtools and awk commands if that's important to you.

[jmartin]

Thanks, thats a good idea. Would I not also need to exclude cases where the 7th column contained '*' as well? So I think it should be:

samtools view foo.bam | awk '{if (($7!="=") && ($7!="*")) print $0}' > different_chromosomes.sam

[dpryan]

If your aligner is outputing singletons then yes, something like that should work.