Support CSI as well as BAI indexes

[peterjc]

The original BAM index format (BAI) has a fixed bin size system, and suffers from a maximum reference sequence length of 512Mbp making it unsuitable for large chromosomes (e.g. many plants, some marsupials). The newer CSI indexing format supports longer references, and also different bin sizes - which ought to help with ultra-high coverage in RNA sequencing or viral samples.

On the C side, htslib and samtools now support both BAI and CSI indexes.

Currently htsjdk only appears to support BAI indexes.

The CSI specification lives at https://github.com/samtools/hts-specs/blob/master/CSIv1.tex and http://samtools.github.io/hts-specs/CSIv1.pdf and http://samtools.github.io/hts-specs/CSIv2.pdf

[jmarshall]

Ignore CSIv2.pdf for the time being. It is a draft from an experimental branch, and CSI v2 may well be different from what that document describes when it finally lands.

[nathanhaigh]

Any progress on this? It would be good to be able to load large (~1Gbp) plant pseudomolecules into IGV using CSI indexes.

[nathanhaigh]

What is required to add CSI support? I may be able to get a colleague to work on this but will need some pointers as to where to start etc.

[jrobinso]

Hey all any updates? I might work on this myself but would like to know the status, if any, first. As this is still open I assume that CSI is not yet supported.

[nathanhaigh]

Not as far as I'm aware. I'm trying to get CSI support in other tools like Sambamba: biod/sambamba#284

[nathanhaigh]

Have you begun working on this yet @jrobinso ?

[jrobinso]

@nathanhaigh Not yet. IGV is taking all my time, very little left to contribute to htsjdk.

[homonecloco]

Are there any updates on this?

[lbergelson]

Htsjdk now supports reading CSI indexes as of 2.19.0