[ensembl-dev] VEP - semicolons in variation database breaking VCF format

[David Parry]

Hello,

I stumbled across a variant after annotating a VCF with the VEP (v76, offline, 
human, GRCh37) that broke the VCF format for downstream annotations. On 
inspection it appears that the 'Existing_variation' feature for this variant 
contains a semicolon.

The variant is:

X       153287314       .       TG      T

And the offending annotation is:

RettBASE_c.*8503delC; 

Having a quick grep for semicolons in the variation cache files there only 
appears to be one other such problem variant (CBS_c.103G>A;129G>A). These were 
present both in GRCh37 and GRCh38. I thought that it might be worth bringing 
to your attention in case other users run into this problem.

Cheers

Dave