Create tabix index from scratch

[holtgrewe]

Hello, where can I find an example on how to create a tabix index from scratch? My use case is that I want to generate a .tbi file for a .vcf.gz file that I'm writing, ideally create the in memory data structures on the fly and then write it out.

A second use case would of course be read .vcf.gz and create the .tbi file from this.

[zaeleus]

The strategy for building an index is to track the start and end virtual positions of each record. Each of these pairs is called a chunk. For tabix, the reference sequence name, record start position, record end position, and chunk are then used by the indexer.

For your first use case, see tabix_write. This shows writing and indexing a BED-like record structure, but it can be applied to VCF as well.

vcf_index shows an example of reading a bgzip-compressed VCF and writing a tabix index for it.

Please also update to noodles 0.54.0 / noodles-csi 0.26.0. While adding the tabix_write example, I discovered a bug (fixed in 599aefa) in the builder, so thank you for asking about indexing!