[Bioperl-l] Database Retrieval

[Sendu Bala]

Sean Davis wrote:
> 
> On 8/7/06 1:53 PM, "Sendu Bala" <bix at sendu.me.uk> wrote:
> 
>> Do you want to go ahead and look into making those classes for
>> accessing the common tables? It's in my plan to make various
>> aspects of genomic data retrieval a strength of bioperl as opposed
>> to a surprising missing link
>> (http://www.bioperl.org/wiki/Getting_Genomic_Sequences); I'll get 
>> to that in a few weeks but if you lay the ground work or better yet
>> complete everything before then that would be great! :)
> 
> So, there is a sketch of what things would look like here:
> 
> http://watson.nci.nih.gov/~sdavis/Bio-DB-UCSC.tar.gz

Thanks for that.


> only includes the refLink and refFlat tables so far, but adding other
> tables is pretty straightforward, as you can see from the code.  I
> would love to hear comments.  Basically, to use, you can do something
> like that shown in the synopsis and output is given below:
> 
> NAME Bio::DB::UCSC - Access UCSC MySQL tables nicely
> 
> SYNOPSIS use Bio::DB::UCSC::RefLink::Manager;
> 
> my $reflinks = Bio::DB::UCSC::RefLink::Manager->get_reflinks( query
> => [ mrnaAcc => {like => 'NM_00002%'}, ], );

I appreciate that this is due to the way Rose::DB works, but is it
possible to hide the SQL nature of what we're doing? Is it possible to
hide even the table names?

Ideally the interface API would survive a complete change in UCSC's
table structures. The implementation would have to change, but user code
would not.

Are you willing to take this on from your outline and develop a set of
more bioperlish modules? Even if you don't have time your contribution 
so far is certainly valuable, so thank you.

I envisage that Bio::DB::UCSC.pm would be the easy-to-use starting 
point, presenting a code interface similar to the UCSC table browsing 
web interface. And while it would implement using various submodules, 
even UCSC.pm would be protected from SQL and table changes.