Query variant filtering

[Will-Tyler]

Overview

This pull request adds support for variant-site filtering in the vcztools query CLI. These changes mimic the filtering expression behavior in bcftools.

This pull request partially addresses #73. I reuse Python components added in #49 to implement these changes.

I also included some small gitignore updates in this pull request.

Testing

I added some unit and validation tests. The query.py coverage was reported as 100%. I'm happy to add more tests if desired.

Usage

vcztools query -f "%POS\n" vcz_test_cache/sample.vcf.vcz -i "POS=112"
112

References

• bcftools query
• bcftools filtering expressions

[tomwhite]

LGTM!

[jeromekelleher]

Very nice, thanks @Will-Tyler!

Just so I'm clear, this works by pulling down the chunks and evaluating for each variant before printout, right? So, if we only printed one site of the VCF we still download and decode all the chunks, and not exit until the entire dataset had been traversed?

If so, there's some performance optimisations we should consider (or at least discuss and reject) where we first pull down the chunks needed to create the variant mask, and then pass the variant mask to the bit that reads chunks. I think we're doing something like this for the region query already?

[jeromekelleher]

Just to be clear, I think we should merge this PR, but open an issue to discuss the performance implications of the implementation.