[DAS2] modification to /type request & response

[Allen Day]

On Tue, 12 Apr 2005, Lincoln Stein wrote:

> Hello,
> 
> The Types request returns the ontology you are using in the "ontology" 
> attribute of the <TYPE> tag.  I think it is fair to insist that if you add 
> a new type to the system you must prepare an ontology that describes it if 
> there isn't one ready-made to inherit from.
> 
> The /type request is only supposed to return those types which are used.  I 
> never expected someone to load the entire sofa database into the thing -- 
> just that subportion of sofa which is actively used to annotate the 
> database objects.

Okay, that makes things easier for the client.  Just a small qualifier
though -- we should return the types used as well as the union of type
graphs *above* those terms.  That way the client is aware it can query
with a generic type (for instance "region") and get back everything.

-Allen

> 
> Lincoln
> 
> On Sunday 10 April 2005 12:27 am, Allen Day wrote:
> > Hi,
> >
> > I'm proposing a change to the way type requests are handled.
> >
> > Background: According to the spec, a genomic feature may be typed by a
> > term from the Sequence Ontology Feature Annotation (sofa), or may be
> > typed by a term from another ontology so long as the term inherits from a
> > sofa term.
> >
> > I have two problems with this.
> >
> > First, there may be other records we want to type -- as a concrete
> > example, let's say we want to type a protein feature as part of
> > protein-DAS, and the type is "alpha helix", which is not in sofa and may
> > be difficult to have inherit from an existing sofa type.  I have other
> > real-world examples from the array-DAS document I'm working on, if anyone
> > cares to see them -- related to biological sample annotation.
> >
> > I'd like to remove the constraint that all types inherit from a type.  It
> > makes sense to keep a constraint like this in place for genome sequence
> > features, but not any other record type.
> >
> > Second, making a /type request is, according to the spec, supposed to
> > return all types known to the server.  If we allow non-sofa-inheriting
> > types into the datasource, the result can be a *very* large document.  As
> > a concrete example, serving up just sofa produces a das2xml 64KB document
> > containing 243 types.  By linear extrapolation, mixing in the Gene
> > Ontology, which contains 18651 types, will produce a document in excess
> > of 4.8MB.  My DAS server actually contains ~15 ontologies that I could
> > conceivably want to annotate with.
> >
> > I'd like to be able to restrict a /type response to a subset of types
> > known by the server, perhaps using the following:
> >
> > * /das/genome/type
> >
> >   returns all types known to the server
> >
> > * /das/genome/type?namespace=sofa.ontology
> >
> >   returns all types known to the server under the "sofa.ontology"
> >   namespace
> >
> > * /das/genome/type?namespaces
> >
> >   returns all type namespaces known to the server
> >
> > -Allen
> > _______________________________________________
> > DAS2 mailing list
> > DAS2 at portal.open-bio.org
> > http://portal.open-bio.org/mailman/listinfo/das2
> 
>