[Open-bio-l] GAME -> BioSQL?

[Hilmar Lapp]

I was actually asking about GAME because we are leaning towards converting every data source into GAME and then only loading GAME into our BioSQL build.

Now your response prompts me to wonder whether I'm missing something. Would people think there is a better suited format or data-flow than going through GAME? (AGAVE? Not widely supported yet ...)

	-hilmar

> -----Original Message-----
> From: David Huen [mailto:smh1008@cus.cam.ac.uk]
> Sent: Friday, June 07, 2002 3:54 AM
> To: Matthew Pocock; Hilmar Lapp
> Cc: OBDA BioSQL (E-mail); GMOD Devel (E-mail)
> Subject: Re: [Open-bio-l] GAME -> BioSQL?
> 
> 
> On Friday 07 Jun 2002 10:59 am, Matthew Pocock wrote:
> > Hilmar Lapp wrote:
> > > The subject basically says it all. Does anyone have a 
> GAME loader for
> > > BioSQL?
> > >
> > > Bioperl has a GAME SeqIO parser (what about BioJava?). Supposedly,
> > > bioperl-db can load Bio::Seq objects into BioSQL 
> (correct?). Is some of
> > > the information originally contained in the GAME input 
> lost along the
> > > way? I.e., if I wanted to, could I round-trip GAME through BioSQL
> > > (assuming that Bioperl GAME SeqIO could write GAME which 
> I think it
> > > can't)?
> > >
> > > 	-hilmar
> >
> > Hi Hilmar,
> >
> > We definitely have a GAME reader. I don't know if we have a 
> GAME writer
> > though. I've had reasonable success round-tripping stuff with the
> > biojava BioSQL driver. Your script would look something like
> > frankenstein's child of game.TestGAME and biosql.UploadFlat 
> both under
> > the biojava-live/demos directory.
> >
> We do have a GAME reader (and I have another that will do 
> more than the 
> original one but not in the same way) but not a GAME writer.  
> The mapping 
> between the Biojava object model and GAME in sufficiently 
> fraught when 
> going from Biojava/BioSQL that I cannot devise a writer that 
> will write 
> GAME in a form that any GAME user could use.  The problem is 
> that GAME as 
> used by Gadfly has a gene model that is rather more concrete than the 
> Biojava semantic free model and interpretation is need to 
> determine which 
> Biojava feature types should be translated into what GAME 
> features.  As 
> GAME is in spec-free mode right now (the spec for me is my 
> interpretation 
> of what I infer from Gadfly data - I think they are the major 
> remaining 
> user of GAME), it's not easy to figure out general solutions.
> 
> Be warned that the GAME reader as it stands uses GAME-1.001 
> coordinates 
> while at least some data out there from Gadfly is in GAME-1.2 
> coordinates 
> which have a different basis!  I do have stuff that does 1.2 
> coordinates 
> too but not in any clean form (ie. suitable for internal use or by a 
> determined hacker only).
> 
> Regards,
> David Huen
> 
> 
> Cambridge, CB2 3UH
> U.K.
>