[ensembl-dev] VariationAdaptor::fetch_all_by_Individual not completely deprecated

[Will McLaren]

Hi John,

I'm not entirely sure I know which methods you are referring to -
neither VariationAdaptor::fetch_all_by_Individual or
CompressedGenotypeAdaptor::fetch_all_by_Individual exist in our
current API.

You are correct, accessing individual genotypes can be slow; this is
because the genotypes are compressed into chunks containing all
genotypes in a 100kb block. Therefore if you are randomly accessing
genotypes from different blocks this can be slow as each time the API
has to retrieve a 100kb block of genotypes and filter out the
appropriate one(s). The schema is optimised for retrieving large
numbers of genotypes along a contiguous genomic region, so if you
configure your scripts to access data in a similar way you should see
speed increases.

This is an issue we are looking into, especially since the number of
genotypes we store is increasing rapidly - the latest release of
Ensembl human contains ~2.5 billion genotypes!

Cheers

Will McLaren
Ensembl Variation

On 22 October 2010 19:26, Ma, Man Chun John <manchunjohn-ma at uiowa.edu> wrote:
> Hi,
>
> I have noticed that CompressedGenotypeAdaptor::fetch_all_by_Individual
> has been deprecated as of August 2007
> (http://listserver.ebi.ac.uk/mailing-lists-archives/ensembl-dev/msg03196
> .html), yet the method VariationAdaptor::fetch_all_by_Individual, which
> directly calls CompressedGenotypeAdaptor::fetch_all_by_Individual, still
> exists to this day. There is no reason to keep a non-functional method
> in Ensembl-Variation, and should be removed.
>
> On the other hand, I have noticed that, probably due to the said
> database restructuring, accessing IndividualGenotypes through the
> Ensembl API is very slow-- are there any way that I can improve the
> speed?
>
> Any help is appreciated!
>
> Cheers,
>
> John MC Ma
> Graduate Assistant
> Kwitek Lab
> Department of Internal Medicine
> 3125E MERF
> 375 Newton Road
> Iowa City IA 52242
>
> _______________________________________________
> Dev mailing list
> Dev at ensembl.org
> http://lists.ensembl.org/mailman/listinfo/dev
>