Alignment and reference sequence from CIGAR and MD

[ingolia]

This is a module that uses CIGAR and MD fields to construct alignments and reconstruct reference sequences from BAM records. I found myself wanting to do this repeatedly, and it isn't actually straightforward. I think these will be generally useful, based on the number of people requesting this feature in various languages.

This code features an MD field parser, a minimal alignment position type that includes only the information directly present in the CIGAR + MD fields, and more "complete" alignment types that use the read sequence to provide complete read and reference sequences directly in the alignment. These are all structured around iterators that generate individual positions, and can easily be collected into a vector if needed. I also added a function to create a bio-type Alignment.

[johanneskoester]

Thanks! I think the API could be simplified, see below.

[y9c]

Hi @johanneskoester

I would like to know if there is a function to get alignment pair now. In other similar htslib bindings, the reference sequence with the query sequence can be fetch in the same time.

For example, https://github.com/blachlylab/dhtslib/blob/11be3debdce9feda903b59ddab6fb737dfd9d3fa/source/dhtslib/sam/record.d#L527-L531

[ArtRand]

Is there any particular blocker for this work? I'd be happy to help get it over the line if necessary.