[BioSQL-l] gbrowse running on biosql

[Lincoln Stein]

Hi Guys,

I've got gbrowse running on top of biosql via a Bio::DasI adaptor.  It took 
about six hours, three of which were to come up to speed on biosql, half an 
hour to write the adaptor layer, and another 2.5 hours to find and fix all 
the latent bugs that a new adaptor exposed in gbrowse!

Basically, the adaptor allows you to type in the accession number of a 
BioEntry, and get a graphical rendition of its feature table, all nicely 
scrollable and zooomable.  Other features, like the various pretty dumpers 
and the restriction map generator, seem to work.  If you're interested in 
trying it, you need to CVS update bioperl-live and the Generic Genome Browser 
CVS (www.gmod.org).  There's a sample configuration file called "biosql.conf" 
that you must modify slightly for your local configuration.

The main limitation of this adaptor is that it uses the 
Bio::DB::BioSQL::BioDatabaseAdaptor module, which is a pure RandomAccessI 
interface.  This means that I can't intelligently fetch features from a 
restricted coordinate range, but instead have to fetch *everything* and then 
filter them for the region of interest.  This limits the usefulness of 
gbrowse to moderate size entries (under a megabase, I guess).  If someone can 
point me to a way to construct virtual entries that contain feature tables 
that have been filtered by coordinate and/or by primary_tag, this would allow 
the interface to scale considerably.

Also, I need volunteers to translate gbrowse's messages file into Spanish, 
Russian, Korean, Japanese and Swedish.  If someone would like to help out 
with this I'll send you a messages file in English, and you just replace the 
English with your favorite language.  It shouldn't take more than 10 minutes.

Lincoln

On Friday 06 September 2002 04:23 pm, Hilmar Lapp wrote:
> I'm thinking about creating a Bio::BioEntry object (no definitive
> name yet) that would basically map to a bare-bones row in the
> Bioentry table in BioSQL (you know that's where I'm coming from
> presently don't you).
>
> BioEntry would be different from PrimarySeq in not having a sequence
> associated with it. Nonetheless, it would have a stable ID and live
> in a namespace; it also has an optional description. Hence,
> technically,
>
> 	Bio::BioEntry is-a Bio::IdentifiableI
> 	Bio::BioEntry is-a Bio::DescribableI
>
> and that's about it.
>
> Note that I'm not describing an ontology term, there is no hierarchy
> or whatever here.
>
> I'm undecided yet whether or not this is really going to help me,
> but as a preliminary poll, if anyone has thoughts on this please
> share.
>
> 	-hilmar

-- 
Lincoln Stein
lstein@cshl.org