gVCF vs VCF variant indexing

[biorococo]

Hi, I see some differences in how variants look when VCF vs gVCF are explored in IGV.

See example in the attached screenshot, positions look exactly in the actual record but index looks different for gVCF. Is this expected?

IGV version is 2.18.4

[jrobinso]

Hi, I would need to see the actual records to comment, but what do you mean by "index"?

[biorococo]

with index what I meant is that the coordinates for the gVCF record look like is spanning two bases vs one base in the VCF.

[biorococo]

Records look like this!

VCF:
chr9 76576987 . AT A 0.00 LowQual T GT:GQ:DP:AD:AF:PL 1/1:0:70:8,5:0.0714:8,0,1

gVCF:
chr9 76576987 . AT A,<NON_REF> 0.00 LowQual T GT:GQ:DP:AD:AF:PL 1/1:0:70:8,5,0:0.0714:8,0,1,990,990,990

[jrobinso]

I think this is normal for a gVCF, in this case the reference is 2 bases and one of the alleles is the generic <NON_REF>, which could be anything, so the variant site covers 2 bases. In the regular VCF there is only 1 alternate allele, and the actual variant can be non-ambiguously interpreted as the deletion of the "T".

@lbergelson knows more about this than I do and might have a comment.