richard apodaca wrote:
> --- On Wed, 9/30/09, Nina Jeliazkova <nina at acad.bg> wrote:
>   
>> I hope I didn't forgot to send you yesterday the link where
>> OpenTox
>> resources are defined: 
>> http://opentox.org/dev/apis/api-1.0/?searchterm=API .
>>     
>
> Nina, thanks - that helps. I missed the API link.
>   
Well, besides me not being clear, perhaps the API pages need more
central place at the site.
> Still, I'm not clear on how these resource relate to one another. For example:
>
> - When using a Compound resource, what subordinate or neighboring Resources should I expect to have access to?
>   

    * Conformers.
    * Features (e.g. identifiers, descriptors, predicted values,
      measured values) .
    * Datasets Compound belongs to.  Datasets will link to Models (as
      training sets for example).  Datasets will link to Validation
      results as well as to Reports. 

Having said that, I could agree this is not clear from the current
documentation, one needs to study schemas and HTTP operations to
retrieve the relations.
> - What's the entry point into the system?
>   

    * Good question. We haven't specified that yet.  According to the
      REST guru, there could be multiple entry points to the system, and
      it is even not necessary all the resources to be reachable (this
      was a bit of surprise to me).
    * Even more important in the context of OpenTox planning to have
      distributed resources, running on different machines, different
      domain names, etc. 

> - What's the most highly-connected resource (and therefor the central concept) in the system?
>   

    * I would say Dataset, but other partners might say Models . And I
      guess a typical end user would say Compound.  All should be fine,
      especially from the point of view of multiple entries ti the system.

> I'm not looking for specific answers to these questions. Instead, I'm looking for tools the OpenTox team can offer or has offered to developers and domain experts to answer these questions for themselves. To stimulate discussion, make sure the right things have been taken into account, and create consensus.
>   
I appreciate your involvement, it really helps to recognize how OpenTox
efforts are seen from outside and hopefully improve.
> An association diagram is not the only game in town, but still helpful. For example:
>
> http://chemcaster.com/rest/resources
>   
This is very clear view indeed.
> Notice how quickly you can: (a) "see" all of the important domain concepts; (b) find the most important concepts by looking for the most connected resource (Registry); (c) discover that you can make Archives from Registries, and Images from Substances.
>
> I'm wondering what possibilities the OpenTox team have explored for doing something like this with OpenTox. In your meetings to hammer out API 1.0, for example, what methods did you use to aid communication and promote consensus about the most important concepts/resources in the toxicity prediction problem domain?
>   
Big white board and talking to each other face to face for couple of
days :)   While this may sound like a joke, this is exactly how we
arrived to API 1.0. Before that, we had several iterations of preparing 
class diagrams (similar to your scheme) and discussing them online,
without really settling on something common.  May be this is the reason
we haven't returned back to recreate the diagram after agreeing on API
1.0.   Currently, next version 1.1 is under development and would be
good to have a visual  clue into the system at the end of the process
(end of October).

Best regards,
Nina
> Best,
> Rich
> ___________________________________
>
> Richard L. Apodaca
>
> http://depth-first.com      Blog
> http://metamolecular.com    Company
>
> _______________________________________________
> Development mailing list
> Development at opentox.org
> http://www.opentox.org/mailman/listinfo/development
>