[Obi-devel] [Fwd: Re: Example Excel file for use with Quick term]
James Malone
James
--
European Bioinformatics Institute,
Wellcome Trust Genome Campus,
Hinxton,
Cambridge, CB10 1SD,
United Kingdom
Tel: + 44 (0) 1223 494 676
Fax: + 44 (0) 1223 492 468
Alan Ruttenberg
when we convene interested parties in August. We need to work this
process out in a way that is sustainable and meets everyone's needs,
or punt and say that it needs to be done outside of the OBI process.
I'll try to give some current thoughts on your questions in line.
On Fri, Jul 17, 2009 at 10:50 AM, James Malone<mal...@ebi.ac.uk> wrote:
>
> Sounds interesting. Since you are writing this now Alan, couple of
> questions from OBI side as this is clearly my main concern. What would be
> the implications to OBI development? How would accessions be assigned
> (given that at the moment we don't assign any accession until release)?
My thought would be to reserve a block of identifiers for the service.
As terms are defined the information is deposited in a place where it
can be picked up by the OBI release process.
> Who would be expected to do curation of these newly created/requested terms
We had talked about making it easy to get identifiers for "defined
classes", as these definitions are managed by the classifier. As we
have been thinking about templates for, e.g. analyte assays, where all
one needs to supply is a CheBI id to get a new logical definition, so
I would expect that we would do the same for antibodies. Part of the
work would be working out what the logical template is for the
antibodies and defining what the minimal set of identifiers that need
to be supplied to the service so a logical definition can be
constructed without intervention. In addition, we already have
curation status markers that indicate the maturity of a term and we
could use that mechanism to indicate the status of submitted terms.
> and what would be the validity of allowing someone to use an OBI accession when
> the class has not been properly curated?
Assuming that we build a logical definition for the antibody, we ought
to be able to have the web service determine whether an antibody is
already in our system. So this becomes a way to create unambiguous
identifiers for antibodies. Ideally other projects, such as text
mining projects, could pick up our identifiers and use them, leading
to the possibility of better integration of data that mentions
antibodies.
> I am certainly for improving and speeding up our ontology creation, without a doubt, but there is an argument
> to say as a reference ontology we are a 'trusted' source for modeling domain
> knowledge.
Indeed. This project will be a lab to see whether it is possible to do
what we've said we want to do, with an appropriate level of quality.
The reason that our grant was funded was to leverage the kind of work
that SCF has been doing by enhancing shared (ala foundry/neurocommons)
resources. So we're explicitly funded to figure out working
compromises between the desire to do things quickly, and the need to
do things in a way which has durable results.
For the antibody work, we don't yet have an idea of how widely the
service will be used. At a minimum the Alzforum, which is a
participant in the project, has a curator to curate their Alzgene
database. Initially submissions to the service would be from him or
one of us working from his work. To expland past that we would explore
just what level of expertise and tools support is needed to provide
adequate information for the service to work, and figure out processes
on the SCF side to ensure that submissions aren't junk.
> Interested to hear more on this though (feel free to bug me :)
You will definitely be bugged :)
-Alan
> ).
------------------------------------------------------------------------------
Enter the BlackBerry Developer Challenge
This is your chance to win up to $100,000 in prizes! For a limited time,
vendors submitting new applications to BlackBerry App World(TM) will have
the opportunity to enter the BlackBerry Developer Challenge. See full prize
details at: http://p.sf.net/sfu/Challenge
_______________________________________________
Obi-devel mailing list
Obi-...@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/obi-devel
James Malone
scenarios; 1) adding cross products of classes we already have in OBI -
this should be fairly safe since we reason over them, 2) adding cross
products for classes that are in OBI and other Foundry ontologies, less
safe as restrictions are not imported into OBI so reasoning over them is
not as much of a safeguard, 3) creating any new cross product class,
most risky and will no doubt require manual curation.
James
--
European Bioinformatics Institute,
Wellcome Trust Genome Campus,
Hinxton,
Cambridge, CB10 1SD,
United Kingdom
Tel: + 44 (0) 1223 494 676
Fax: + 44 (0) 1223 492 468
Philippe Rocca-Serra
From the limited experience I have gathered writing a perl script to
add qtt relying on a spreadsheet template,
the issue of making sure that new OBI identifiers can be issues by a
central authority is key.
in the absence of this central authority, since we are working in
branch, we need to decide if we allow anyone to run the procedure as
needed or if we assign the task to a x-product/qtt people to reduce the
risk of identifier clash.
Then , when it comes to case-2, is this a matter of modifying the mireot
mechanism and extend it
Cheers
P
--
Philippe Rocca-Serra, PhD
Technical Coordinator
www.ebi.ac.uk/net-project
The European Bioinformatics Institute email: ro...@ebi.ac.uk
EMBL Outstation - Hinxton direct: +44 (0)1223 492 553
Wellcome Trust Genome Campus fax: +44 (0)1223 492 620
Cambridge CB10 1SD, UK room: A3-141
--
Alan Ruttenberg
On Fri, Jul 17, 2009 at 12:50 PM, Philippe Rocca-Serra<ro...@ebi.ac.uk> wrote:
>
> From the limited experience I have gathered writing a perl script to add qtt
> relying on a spreadsheet template,
> the issue of making sure that new OBI identifiers can be issues by a central
> authority is key.
> in the absence of this central authority, since we are working in branch, we
> need to decide if we allow anyone to run the procedure as needed or if we
> assign the task to a x-product/qtt people to reduce the risk of identifier
> clash.
>
To manage this I suggest we could allocate identifiers in batches to
people who want to do this. This minimizes the need to hit a central
authority frequently - you just need to do that when you run out of
the ids in your batch.
Bjoern Peters
- QTT (quick term template) for the creation of defined classes following a restricted design pattern, which should ensure that validity and duplicity can be checked for automatically
- Quick ID for more generic submissions of new classes, which will necessarily be subject to manual review, but where establishing an ID early on could speed up adoption and turn-around time.
In James list, QTT is a subset 1+2.
It is great to hear that Alan has resources for a developer and the Protege team is working on the spreadsheet importer! Philippe has put together a wish list of software functionality based on the QTT work, and it would be great if a solution to these could be produced jointly or at least interoperable.
- Bjoern
Trish Whetzel
[I hate that google remove the default reply to all setting]
To manage this I suggest we could allocate identifiers in batches to
On Fri, Jul 17, 2009 at 12:50 PM, Philippe Rocca-Serra<ro...@ebi.ac.uk> wrote:
>
> From the limited experience I have gathered writing a perl script to add qtt
> relying on a spreadsheet template,
> the issue of making sure that new OBI identifiers can be issues by a central
> authority is key.
> in the absence of this central authority, since we are working in branch, we
> need to decide if we allow anyone to run the procedure as needed or if we
> assign the task to a x-product/qtt people to reduce the risk of identifier
> clash.
>
people who want to do this. This minimizes the need to hit a central
authority frequently - you just need to do that when you run out of
the ids in your batch.
Identifiers had been allocated in batch in the past, but then that practice was reverted - how will this set-up be different?
Trish