Return pairs in ReviewConsensusVariants

[SPPearce]

When applied to paired end data, ReviewConsensusVariants returns only the reads that contain the mismatch with the reference, not the read pairs.

Is it possible to add an option to return the pairs as well?

[nh13]

That's a tough one at first blush, since if we have a mate that's on a different chromosome, we'd have to retrieve it as well, and depending on the output BAM sort, that could get complicated. A work around would be to collect the read names from the output of ReviewConsensusVariants, then re-filter the input of ReviewConsensusVariants using samtools view --qname-file. It's inefficient, but sort-of how we'd have to do it anyways.

[SPPearce]

Yeah, I know that dealing with pairs is a pain, particularly when you start thinking about cross chromosomes or supplementary/secondary alignments.

[nh13]

I could see writing a class that implements a SamWriter that takes (1) a writer to write the final records to, (2) a reader with all reads. When you write records to it, it keeps a list (hash) of all the read names it sees, and when it's closed, it goes through the input (2) and filters for all the records with a matching name and writes it to (1).

[nh13]

An example would be #1037, but I'd want to flesh it out and wire it in.