vcf: Define Number value P about hts-specs HOT 6 OPEN

Would P even make sense for INFO column? I think maybe P should only be valid for FORMAT fields where a sample has a defined ploidy so it should just be defined in the Genotype section. Some samples may be triploid some may be diploid for the same reference genome. Specifically, P is the number of alleles in the FORMAT/GT field.

On a related note, should this FORMAT field have Number=P rather than Number=1?:
##FORMAT=<ID=HAP,Number=1,Type=Integer,Description="Unique haplotype identifier">
and can this be used to link phased haplotypes from multiple records? I'm thinking of using this FORMAT/HAP field to identify star alleles for PGx with multiple VCF records across a gene.

from hts-specs.

Note that this P is used in the current VCFv4.4 specification for the PSL, PSO, and PSQ genotype fields (though these may not yet have seen widespread use), and has been suggested as being appropriate for some of LAA/LAD/LGT (cf #434).

from hts-specs.

Added FORMAT P definition to VCF 4.5. No INFO P though as VCF make no assumptions about ploidy outside of GL. Use a . or a fixed number (e.g. Number=2 for diploid) if the ploidy is known.

from hts-specs.

TODO: errata 4.4 to change P to . since it wasn't properly defined.

from hts-specs.

…or add the definition of P to the 4.4 spec. Pros and cons on both sides; IMHO it's best for the 4.4 and 4.5.draft specs to say the same thing for PSL/PSO/PSQ.

Possibly also add a reference to this to the errata or changelog sections.

from hts-specs.

Let's go with defining Number=P and explicitly listing it as an errata in Section 7.

from hts-specs.