[ensembl-dev] Variant Consequence Predictor

Will McLaren wm2 at ebi.ac.uk
Fri Feb 25 11:36:00 GMT 2011

Hi Stuart,

> Hi Will,
> Thanks for your response. So now I have a problem because it seems what I
> have been doing is possibly invalid:
> I have downloaded the tables I need from the DB
> homo_sapiens_variation_61_37f (variation_feature, transcript_variation,
> allele etc). I then link the variation_feature to the transcript_variation
> table to get the peptide change and to the allele table to get the relevant
> alleles for a given population. But what you are saying is that the peptide
> change in the transcript_variation table doesn't necessarily equate to the
> alleles in the allele table for any one given variation_feature?

This approach is valid, but unfortunately due to inconsistencies in
dbSNP which we haven't yet fully figured out how to resolve there are
situations such as this one where the alleles present in the allele
table do not correspond to the alleles present in variation_feature's
allele_string field. It is the allele_string field that we use to
calculate the consequences in transcript_variation.

Fortunately these odd situations are pretty rare, so in the vast
majority of cases your approach will work just fine.

We are currently in the process of overhauling the system such that
consequences will be linked to particular alleles. We will also be
taking a look at how we can resolve situations where alleles appear in
allele but not in variation_feature and vice versa.

Hope this helps


> Cheers
> Stuart
>> Hope this helps
>> Will

More information about the Dev mailing list