[ensembl-dev] Ensembl VEP reports incorrect HGVSc for chr X variants

[Pär Engström]

Hi,

I have come across an issue with Ensembl VEP. I have used the web version of VEP for GRCh37, but the same problem seems to occur with the web version for GRCh38.

VEP has an option to report HGVS strings. This works fine for variants on the autosomes, but not for X chromosome variants. For those, the cDNA HGVS strings are frequently incorrect. For example, try the two following variants (VCF format):

X 31222203 1 A G . .
X 31697653 2 G C . .

These variants fall within the coding sequence of a minus strand gene, so the first should be reported as "T>C" in the HGVSc strings. However, VEP reports it as "A>C". The problem seems to be specific to the HGVSc field, because the Codons field is correct (TTT/CTT). The HGVSp field also seems to be correct.

For the second variant, the error is somewhat different: here the HGVSc string is empty for most Ensembl transcripts, although the Codons field lists the correct change (CGG/GGG).

The problem seems to occur with GRCh38 as well (i.e. current Ensembl). I tried the first variant listed above. The result was different compared to the GRCh37 version, but still wrong. Now the change was indicated as N>C in the HGVSc strings. The variant in GRCh38 coords:

X 31204086 1 A G . .

I have processed quite a few variants and not seen this problem for any autosomal variant, but for many chr X variants.

Kind regards,

Pär