genotype_count_spec is inconsistent with VCF spec

[timothymillar]

Just a minor inconsistency I noted given that we seem to follow the VCF spec in most cases. The genotype_count_spec states "the 3 columns contain heterozygous, homozygous reference, and homozygous alternate counts (in that order)". The VCF genotype ordering would typically be homozygous reference, heterozygous, homozygous alternate for a diploid biallelic variant as described for the "GL" field in the spec.

[timothymillar]

Just to add to that, If we were to change the ordering to match VCF ordering then this variable could easily be generalized to multi-allelic variants and autopolyploids.

[tomwhite]

+1 to introducing a generalized ordering like GL.

It looks like the genotype_count variable is only used in hardy_weinberg_test - and it's optional. Does HWE generalize to multi-allelic/autopolyploids? If not, then we might leave HWE as it is, and rename its variable to hwe_genotype_count_spec or similar. If it does then we might want to consider changing it in HWE too, but may need to think about backwards compatibility (as it's optional it's possible it's not being used anywhere). I don't know why that order was used in HWE (@eric-czech might do).

BTW here is the relevant test: https://github.com/pystatgen/sgkit/blob/68b983c51c2c7e08914d64f83cd7b3fd05616f1a/sgkit/tests/test_hwe.py#L139

[timothymillar]

Does HWE generalize to multi-allelic/autopolyploids?

Yes in general although I'm not familiar with implementation details (something on my long term todo list).

[timothymillar]

@tomwhite I'm not sure how urgently you want to release 0.6.0 but I can try sort this tomorrow.

[tomwhite]

Thanks Tim, that would be great - I'll try to release next week.