Repository navigation
Downstream plugin inconsistently reports a leading wild type (unmutated) amino acid #342
Description
Activity
- changed the title
[-]Downstream plugin inconsistently reports a leading wild type amino acid[/-][+]Downstream plugin inconsistently reports a leading wild type (unmutated) amino acid[/+]on Aug 5, 2020 Hi @susannasiebert,
the Downstream plugin never reports a leading wild type amino acid. It always reports the amino acid of the first codon which is affected by the variant allele.
Since release 100 we introduced the option of shifting variant alleles in repetitive regions. The new functionality is switched off by default except for the HGVS calculation. The user from griffithlab/pVACtools#576 who uses VEP release 100 and is also asking for HGVS gets a different Downstream amino acid sequence than when using VEP release 95 or when using VEP release 100 without hgvs option.
And the difference is due to the shifted position of the variant allele in release 100.But I need to emphasise that I'm not expecting that when --hgvs is used that it has an effect on the calculations in the Downstream plugin. This looks like a bug to me and we need to investigate further. I will keep you updated here with any progress we make on the issue. For now I would recommend that the user shouldn't use --hgvs when annotating the input data which is then used by your tool.
Best wishes,
AnjaPlease see griffithlab/pVACtools#576 (comment) for example results from running VEP 100 with and without the
--hgvsflag. As suspected, not using the--hgvsflag does result in a leading wildtype amino acid in the Downstream protein prediction.Thank you for letting us know. We are still working on a fix and will let you know as soon as we have updated the VEP code.
Thank you for looking into it. On a (marginally) related note: would it be possible to output the version of a plugin in the plugin's VCF header entry? This would make it possible for our downstream tools to ensure that the new/fixed plugin was used during annotation.
@at7 I was wondering if there had been any updates on this issue?
Hi @susannasiebert,
we are still working on a bug fix and hope to release the fix with the new Ensembl version 102 which we are planning to release by the end of October. I will keep you updated here about any progress.Best regards,
AnjaWonderful. Thank you!
Hi Susanna,
I would like to give you an update for this issue. We have been working on a bug fix but we will still need some more time for testing. I will give you another update this week with a more concrete time for when we can provide the fix.
Thank you very much for your patience.
Anja@at7 we were able to resolve this issue on our end by switching to a different plugin we wrote ourselves called
Frameshiftwhich reports the full mutated transcript protein sequence for frameshift mutations.Is there a process to submit new plugins to VEP for consideration?
We've encountered situations were sometimes the Downstream plugin will report one leading wildtype amino acid and sometimes it does not. For example this user (griffithlab/pVACtools#576) using the GRCh37 cache reports result that do not include a leading wild type amino acid while this user's prediction on GRCh38 (griffithlab/pVACtools#484) does return a leading wild type amino acid. I'm not sure if the different reference builds are indeed the problem but in our variant prediction pipelines using GRCh38 and VEP95 we are also encountering the leading wildtype amino acid.
Is this intentional and if so, is there a heuristic to determine when a Downstream prediction includes this leading wild type amino acid?