PP field logic in phase_rare VCF output

[creepyorange]

Hello!
Thank you for developing and maintaining this incredibly useful tool.
I am using shapeit5 for phasing whole-genome sequencing data from a large cohort, with a particular focus on the phasing of rare variants. I am trying to understand the PP (phasing confidence) field in the output VCF.
As far as I understood the PP field can often be missing (.), and I want to figure out the precise conditions under which a non-missing PP value is assigned.
My initial assumption was that a non-missing PP would require a heterozygous genotype (e.g., 0|1 or 1|0). However, some homozygous reference genotypes (0|0) have a PP value of 1 while other homozygous reference genotypes have a missing (.) PP field.

Could you please clarify the logic behind the assignment of the PP tag?

shapeit version: 5.1.1

Command used: phase_rare_static --input {input.to_phase} --scaffold {input.scaffold} --map {params.gmap}
--input-region {wildcards.interval} --scaffold-region {wildcards.interval} --output {output.rare_chunk} --thread {threads}

[srubinacci]

If a homozygous genotype has a PP field assigned, it means that the genotype was imputed — in other words, it was missing in the original file.
When the PP value is missing (“.”), it simply indicates that the phasing probability is 1.
If the PP field is not assigned at all for a given variant, it means that the variant is common and thus not subject to rare variant phasing.