Why does Ensembl VEP provide HGVSg for some variants and not others?

Question

[Jeremy Leipzig]

docker run \
            -u 1000:1000 \
            -v $HOME/vep_data:/data \
            -v $HOME/vep_data:/.vep \
            -v $PWD:/opt/foo ensemblorg/ensembl-vep \
            vep \
            -i /opt/foo/{input} \
            -o /opt/foo/{output} \
            --cache \
            --force_overwrite \
            --fork {threads}  \
            --format vcf \
            --buffer_size 5000 \
            --terms SO \
            --symbol \
            --ccds \
            --variant_class \
            --hgvs \
            --hgvsg \
            --force \
            --dont_skip \
            --no_stats \
            --pick_allele \
            --vcf \
            --show_ref_allele

no HGVSg:

chr22   10510044        .       G       T       .       .       CSQ=T|intergenic_variant|MODIFIER||||||||||||||||G||||SNV|||||| GT      0/1

2 base pairs later, HGVSg:

chr22   10510046        .       A       T       .       .       CSQ=T|intergenic_variant|MODIFIER||||||||||||||||A||||SNV|||||chr22:g.10510046G>T|      GT      0/1

[LauferVA]

3 quick questions:

1) are you working with a single sample or multisample VCF?

2) what genomic build? (ill assume GRCh38.p14)

3) do you have strand information?

Pending that, best guess:

T is reference for GRCh38 - GRCh38.p14 for 22:10510044, while G is reference for 22:10510046. Therefore it may simply owe to designation as a variant or not... no matter the strand considered for 22:100510046, it has to be a variant (viz. A/T; T/A on flip strand); versus 22:10510044 (G/T; C/A) could ostensibly be a variant or could be reference.

If you provide 15 of each (vars with and without anno, could confirm/disconfirm this guess with high confidence)

i don't think it has to do with presence/absence in a given database, after cursory cross-checks. if none of this pans out, would probably ask Emily Ensembl

[GenoMax]

FYI: Emily no longer works for/at Ensembl. That is the reason she has taken out _ensembl from her profile.

Ben_Ensembl may perhaps have some input.

[LauferVA]

GenoMax thank you for clarifying that - end of an era!

[Jeremy Leipzig]

you're right it's an off-by-one-error on my part. thanks

[LauferVA]

I have a couple thoughts on this - first, though, would like to confirm that you are on GRCh38?

[Jeremy Leipzig]

yes this should be a reproducible example using just those two lines and the latest vep and database on GRCh38

[LauferVA]

copy that.

also granted your background i expected nothing less :-)