[MOBY-l] MOBY Central database schema

[Jason E. Stewart]

"Mark Wilkinson" <mwilkinson at gene.pbi.nrc.ca> writes:

> Those last two are the sticking points.  At the moment the objects
> and services are in flat files, but I think we want them in tables
> which represent their ontology.  The GO database schema seems
> unnecessarily heavy, but on the other hand we get the tools for
> free.  

As I just found out, though - none of their tools currently support
ontologies that are not GO. I struggled like made to get things
working and finally convinced Dag-Edit to put things in my Postgres
DB, but the Perl API doesn't recognize things. Chris Mungall said that
he will be committing a bunch of stuff for the Perl API after next
weeks GO meeting that specifically deal with support for non-GO
ontologies. 

But for now, I don't want to expend any more time on
technicalities. We need two things from these ontologies:
1) Lookup a term's information based on ID (e.g. the term name and
   definiation)
2) What terms are children of that term (for looking up a service
   based on the service type - we want all child service types as
   well) 

If you have a good feeling for how to handle a tree (let's stick with
trees for now and not DAGs) in an SQL DB, then by all means go
ahead. My simplistic idea was simply to allow Ontology entries to have
a single parent, and then to have a second de-noramlized table that
listed all the children of any given node.

> For the moment I'll just make a set of simple tables with a third
> table "term2term" like they have in GO... or something like
> that... just to get us started.

I don't know their schema at all yet, what does the term2term table do?

> After that, we need some actual XSD specifications for Objects.  Jason,
> are you building some of the ones that we talked about at the
> meeting?

I haven't yet. My familiarity with W3C XML Schema is minimal. Also,
what Perl tools will you use to parse the XML files - I know that
XML::Xerces will handle schemas (I maintain it), but are there others
that are schema-aware yet?

> I'll also take a stab at a few on Monday just to make sure that I'm
> generating the WSDL document correctly.  I can't think of a better
> way to store these than in flat-files somewhere on MOBY-Central,
> since they are going to be imported and modified during the creation
> of the WSDL document... does anyone have a better suggestion?  In
> any case, that will work for the time being.

I think storing them as text in the DB should work. We can have a
separate WSDL table.

> I'm going to go out on a limb and say that I hope to have
> MOBY-Central available on a public server by the following
> Monday.... but please don't hang me by my toes if I fail!

Hey, I like ambitious...

> Midori, Jason - I've made a few requests to get you added as open-bio
> developers, but I'm not getting any response.  I've just sent the
> request to someone else to see if I get an answer.  Sorry it is taking
> so long!

I just called up Andrew Dalke and I'll see if he can help get things
rolling.

> Best wishes all, and have a great weekend!!  I'm off to celebrate my
> 34th  :-)   don't wait up...

Ja ma han leva, ja ma han leva, ja ma han leva ut i hundrade ar!

(swedish b-day wishes ;-)

jas.