On 12 Aug 2010, at 12:16, Thomas Down wrote:
>
> A few other questions (don't really have "preferred" answers to any of these,
> just trying to test the boundaries):
>
> 1. What do you expect a server to do if it sees a CS URI that it
> hasn't seen before?
I assume you mean an alignment server receiving such a URI as a parameter? If
so I think there is only one realistic thing it can do right now, which is
return nothing (i.e. no alignments). It is not an authority on which coordinate
systems are real and which aren't, it just knows about the ones it is using. In
essence it simply has no data for the requested subject because it doesn't
recognise the subject. We could invent error situations for things like this,
but that would make DAS more complex for little gain IMO..
> 2. If my organization has sequenced a new genome and is running some
> internal DAS stuff on that while we finish annotating, etc., what URI do we
> use for the coordinate system?
>
> 3. If my organization is running an internal mirror of the central
> DAS registry, would I mirror the CS URIs
> ("http://das.bigpharma.com/dasregistry/coordsys/CS_DS40/";). Still point to
> dasregistry.org? Something else?
As Jon says, I think it's easier just to create the coordinate system in the
registry. At the moment you need admin privileges to do that via the interface,
but I implemented a programmatic version of this wherein the registry accepts
POSTed coordinate system XML (sans URI) and replies with the same (with URI
included). It is restricted to allow you to create new authority/versions, but
not new "segment types": so you can create "GRCh_38,Chromosome,Homo sapiens"
but not "GRCh_37,wibble,Homo sapiens". We were a bit wary of opening up the
ability to do the latter in fear of ending up with different spellings of the
same thing, or other similar issues.
I suppose the same thing could be made available to registered users in the
interface (possibly at the same time reworking it to use the web services
underneath).
> Thomas.
_______________________________________________
DAS mailing list
[email protected]
http://lists.open-bio.org/mailman/listinfo/das
