[MOBY-l] thinking about provision info and authoritative service providers

[Lincoln Stein]

The LSID view is that there is an authority for a namespace.  That is, only 
NCBI is allowed to give a GenBank ID to a sequence entry.  We need to 
distinguish between who has the right to associate a recognized name with a 
data object, and who has the right to perform computations on it and to 
aggregate objects in various ways.


Lincoln

On Friday 25 October 2002 01:51 pm, Mark Wilkinson wrote:
> Hi everyone,
>
> There are a bunch of thoughts crystallizing in my mind at the moment
> around issues of provision information and versioning and such.  It has
> come up repeatedly as I give the MOBY presentation in various places,
> and is clearly something that scientists are very concerned about right
> from the get go.  In addition, the good folks at myGrid have seen it as
> sufficiently important that it is a major part of their entire platform,
> whereas we have largely "put off" the problem.
>
> This was all brought to the surface yesterday during a telephone
> conversation with Lincoln, Damian, Andrew, et. al..  The issue came up
> of "who gets to serve", for example, GO_Terms.  It wasn't phrased that
> way, of course, but that was in some ways the essense of the issue.
>
> Given that we put no restrictions on who may provide whatever service,
> it is possible (likely!) that many services will be providing outdated
> information. Lincoln suggested that we perhaps flag certain bodies as
> being "authoritative". A few tweaks to the MOBY Central database and API
> would be sufficient to store this information and also limit requests
> to, for example, "authoritative services only".
>
> ...but this opens up some questions in my mind... apart from the
> political jockying about who gets to *be* authoritative and on whose
> head that decision is going to fall ;-)
>
> I was thinking about the use of the word "authority" in the MOBY versus
> LSID world.  In both cases, we represent the "authority" as a URI
> effectively naming an organization.  However, in LSID the authority is
> (in my interpretation of the spec) tightly linked to a namespace, while
> in the MOBY world the authority is tightly linked to a service.  So LSID
> folks might intuitively believe that the authoritative server for the
> Genbank namespace is NCBI, while in MOBY this is not always going to be
> the case... for example, if the Gene Ontology consortium decides to set
> up a service that takes in GO_Terms and spits out GenBank records of
> sequences in GO that are annotated to that term... then they are,
> clearly, the authoritative providers of GenBank in that circumstance,
> since they have the most up-to-date information on the
> **transformation** that they are providing.  This leads us to a somewhat
> peculiar situation where we could ask "who is the authoritative provider
> of Genbank records" and both NCBI and GO appear as the answer.  Of
> course, the truth is that NCBI is the authoritative transformer of
> GenBank GI's or Accessions into GenBank records, while GO is the
> authoritative transformer of GO/ID's into Genbank records... but it is
> still a bit peculiar.  I don't know if it is going to be problematic in
> the end, but it is something to keep in mind as we move towards adoping
> authoritative services....
>
> The other thing that directly relates to this is versioning information.
>   Whether you are using the authoritative service or not, you still want
> to know the version of the resources (both data and software) that are
> being provided to you.  Again, because our services (under the current
> MOBY architecture) are generally unrestricted, this might be confusing,
> but at least we seem to have a place to fit this information.  The
> underused <MOBY/> envelope isa natural place to stick this
> information... perhaps as attributes of the MOBY envelope
>
> <MOBY authority="geneontology.org" GenBank="version xx"
> GeneOntology="version yyy" authoritative="yes" Date="12-12-02 13:05:02">
>
> 	<Sequence namespace="Genbank/GI" id="123456">
> 		...
> 		...
> 	</Sequence>
> 	<Sequence namespace="Genbank/GI" id="789012">
> 		...
> 		...
> 	</Sequence>
> </MOBY>
>
>
> I realize that the vocabularies for the resources aren't controlled, but
> we can hopefully find a way to deal with that...  nevertheless, is that
> just too ugly, or do you think that is sufficient to accomplish what is
> needed?  It certainly is a simple solution...
>
> Philip/Carol/Robert - how do you represent this data in myGrid?  Have
> you found that there a need to be more complicated than that?
>
> M

-- 
========================================================================
Lincoln D. Stein                           Cold Spring Harbor Laboratory
lstein at cshl.org			                  Cold Spring Harbor, NY
========================================================================