[ensembl-dev] Variant Effect Predictor "check_ref" feature

Will McLaren wm2 at ebi.ac.uk
Tue Aug 2 14:56:39 BST 2011


Hi Mark,

We actually already precompute every possible Polyphen and SIFT score
already - they are combined using the Perl API method
get_condel_prediction in Bio::EnsEMBL::Variation::Utils::Condel to
produce the score you see.

If you don't mind a bit of Perl coding, you will probably find it
quicker to download our prebuilt tables and calculate using the API
from there, rather than using the VEP to do this.

The tables you need are:

protein_info
protein_position
polyphen_prediction
sift_prediction

You can get them from here:

ftp://ftp.ensembl.org/pub/release-63/mysql/homo_sapiens_variation_63_37/

You could probably even extract the info you need directly from these
files, without loading them into a Mysql instance.

Regarding strand, for VCF the script assumes the variant is on the
forward strand (the alleles are flipped when analysing a transcript on
the reverse strand). In the default Ensembl variant format, you can
specify strand in the 5th column of the input (using "+" or "-").

Cheers

Will

On 2 August 2011 14:19, Mark Aquino <aquino.m at husky.neu.edu> wrote:
> I'm trying to actually pre-compute condel scores for every codon possibility in the coding exome by supplying the VEP each position, a ref codon, and a var codon. Anyways, in that case I'll try to use the API to generate the reference codons at each position in the script I used to generate the input file.  Another question, how does your script determine what strand the variation is occurring on? I can't seem to figure that out.
>
>
> On Aug 2, 2011, at 4:43 AM, Will McLaren wrote:
>
>> Hi Mark,
>>
>> At the moment it just checks, throws a warning and skips that line of input.
>>
>> If you would find it useful for the script to change the reference
>> allele, this is an option I can look into adding in the next release.
>>
>> However, you are then left with the issue of whether you want to
>> discard the reference allele you supplied, or to submit it as an
>> additional variant allele. There's also the case where you have
>> submitted, for example, ref A var G and the reference is actually G.
>>
>> What is the use case for this?
>>
>> Will
>>
>> On 1 August 2011 22:30, Mark Aquino <aquino.m at husky.neu.edu> wrote:
>>> Hey Will,
>>>
>>> Does adding the check_ref just throw a warning if the supplied allele doesn't match the reference on Ensembl or does it (could it?) change the incorrect ref allele to the Ensembl ref allele?
>>>
>>> Best Regards,
>>> Mark
>>>
>
>




More information about the Dev mailing list