[ensembl-dev] method to fetch 5' and /or 3' from a feature- ensembl variation

[Kieron Taylor]

Hi Nathalie,

On 04/08/2014 11:40, Nathalie Conte wrote:
> HI Kieron, Thanks - At the moment I am using the  get_nearest_Gene()
> from core-
> http://www.ensembl.org/info/docs/Doxygen/core-api/classBio_1_1EnsEMBL_1_1Feature.html#a8a7c8af129910c4daa088b87171b86a6
>
>I receive a warning that this method is deprecated:
> ------------------ DEPRECATED --------------------- Deprecated method
> call in file
> /net/isilon5/ma/home/nconte/src/ensembl/modules/Bio/EnsEMBL/Feature.pm
> line 1434. Method
> Bio::EnsEMBL::DBSQL::GeneAdaptor::fetch_nearest_Gene_by_Feature is
> deprecated. fetch_nearest_Gene_by_Feature() is a deprecated method.
> Strand and stream implementations are unreliable. Please use with c
> aution Ensembl API version = 75

The method is deprecated because it produces rather unpredictable output 
in certain situations, but it will continue to function as it did before 
until it is properly retired. Caveat emptor.

> I have several questions please  - When the next method you mentioned
> : feature/getnearestgene will be available? and will it fetch nearest
> 3' and 5' at the same time? Will we be able to narrow the search to
> only protein coding genes rather than all genes?

The new code should be included in our October release (77). The date 
for 77 may shift around a bit as we are only just about to release 76, 
so I can't give you a better estimate. Once we have finalised its 
behaviour and tested it to destruction, it can be used from the feature 
branch if you are prepared to accept the baggage of running pre-release 
code. I can notify you when this happens if you would like.

The new method is intended to scan in whichever directions the user 
chooses, e.g. up and downstream by 1000 bp, or only downstream on a 
particular strand, overlapping, non-overlapping and so on. I have no 
intention of restricting the results to coding genes, as the method is 
generic to all features, but I can help you with the filtering process. 
The complexity of the method will necessitate a number of convenience 
methods that will cover some or maybe all of what you are asking for. 
Can you confirm whether you are looking for absolute nearest or many 
nearby genes?

>
> Thanks Nathalie
>
>
> On 21 Jul 2014, at 14:07, Kieron Taylor <ktaylor at ebi.ac.uk> wrote:
>
>> Hi Nathalie,
>>
>> We are working on a method to support this kind of enquiry, but it
>> turns out to be quite finely nuanced to find the nearest feature
>> when it overlaps others. I hope to get it included in release 77.
>>
>> If you're curious, take a look at the feature/getnearestgene branch
>> on Github. This code is functional but I don't recommend using it
>> until we release it properly, as it will continue to evolve.
>>
>>
>> Regards,
>>
>> Kieron Taylor Ensembl Core
>>
>>
>> On 18/07/2014 16:27, Nathalie Conte wrote:
>>> Hi Dev team, Any method in ensembl perl API to fetch nearest gene
>>> in 5' and 3' from a feature (ie variation object?)? thanks
>>> Nathalie
>>


Regards,

Kieron