First login?
Lost password?
 
 
cellml-discussion AT lists.cellml.org 
CellML Discussion List

Text archives Help


[cellml-discussion] Suggestions regarding search and browse utility for CellML Repository


Chronological Thread 
  • From: ely.matos at ufjf.edu.br (Ely Edison Matos)
  • Subject: [cellml-discussion] Suggestions regarding search and browse utility for CellML Repository
  • Date: Sat, 3 May 2008 00:11:26 -0300
Hello all,
James, my Msc. dissertation, recently concluded, was about the use of
ontologies to enrich semantically the CellML Models. I've proposed a initial
ontology represented in OWL-DL and using SWRL, called CelO (Cell Component
Ontology) with 3 "sub-ontologies": one to represent SI units and quantities,
other to represent the biological domain terms and another to represent the
model.

My proposal was exactly to have "atomic components" (models with a single
component or, at least, a single "interface" - when a component "encapsules"
other components and only it is visible to external world). Each "atomic
model" owns a interface (input/output variables) semantically annoted to let
one to make "semantic queries" about models (using SPARQL). This let me to
do questions as "which are the components related with SodiumIonChannel?",
taked in account that this "knowledge" is extracted from the names of
variables in CellML Model (like "I_Na") and some inferences (like "I" is the
symbol of "ampere" that measures "eletric_current", then this variable is
associated with eletric_current; as it is defined inside the component
"mebrane" and "Na" is the symbol of "Sodium", it can be associated with
.SodiumIonChannel). This is a very simple example.
.
Each "CelO Model" has a reference to the correspondent CellML Model, to let
compose "atomic components" and create a "composed model", with components
of diverse models (as a rough example, a SodiumChannel of Hodking_Huxley
inside a Noble model). The CelO Models are stored in a relational database
and inference engines can be used to do queries.

I've implemented a "framework" (as web service) that offers all this
services (registry, query, "execute", compose models).

This work is very incipient, given my "almost null" knowlegde about
intrinsics of biological domain. As my Master Program is about Computational
Modeling, my focus was strongly directed to the "modelling" part, not the
"biological" part. My objective was provide a "high level" view of CellML
Models.

Unfortunately my dissertation is written only in brazilian portuguese. I've
submmited some papers but I didn't get success. However, if there is some
curiosity I can try to translate the main ideas.

Sorry by intromission and the bad english,

Ely Matos
Computational Modeling Master Program
UFJF/Brazil


----- Original Message -----
From: "James Lawson" <j.lawson at auckland.ac.nz>
To: "CellML Discussion List" <cellml-discussion at cellml.org>
Sent: Wednesday, April 30, 2008 11:04 PM
Subject: Re: [cellml-discussion] Suggestions regarding search and browse
utility for CellML Repository


> Hi Abhishek,
>
> Good to hear from you, you make some very good points. I apologise if
> someone has already raised these points, I've just come back after being
> away for a couple of weeks and I'm going through my many unanswered
> emails.
>
> So at the moment, one of the things I am thinking a lot about is
> ontological annotation of CellML models. I've just been to visit Neil
> Wipat's lab at the University of Newcastle, and I talked quite a lot
> with him and members of his group about ontologies and how they might be
> of use to the CellML community. So one of the things that the new CellML
> repository will handle much better than the current one is the manner in
> which a user can view models and their interrelationships, whether these
> are based directly on CellML relationships such as import based
> modularity or whether they are based on relationships inferred from
> ontological tags.
>
> I'll definitely have a look at MeSH - what we're planning to do (I
> believe) is let the CellML-Bio Ontology import as much as possible from
> other ontologies, and let Sarala's ontological framework stand as our
> contribution. Where we conclude that there is no ontology that covers
> what we are trying to do in a satisfactory manner, we'll create our own.
> I can see this happening in the area of synthetic biology, which is only
> just starting to be represented by formats like CellML and SBML. So a
> tree-like browsing structure as constructed from MeSH tags or similar
> could be one of a number of different ways of viewing the models. You
> could also do searching by terms, or construct trees according to terms.
> For example, you could have 'species' as the top level and 'organ' as a
> lower level, and 'cell type' as the bottom. So then you'd get a tree
> sorted first by species, then by organ type, say heart, pancreas, liver,
> etc. and then by cell type, so within the heart: myocardium, epicardium,
> etc. etc.
>
> The possibilities for how we display fully tagged models are huge, so we
> do need to form some kind of system of constraints - or at least provide
> a set of default viewing architectures based on what workflows we
> anticipate for users.
>
> When we decompose models into modular elements, this kind of tagging is
> going to be absolutely crucial for model reusability and for the process
> of figuring out what module you want and where to get it.
>
> As far as the technology for web services and programmatic access to the
> repository database, that isn't something we've thought about much until
> now, but on our travels Mike and I have realised that this is really
> quite important. In fact, it was often one of the first questions I was
> asked when I was discussing the CellML repository. A lot of people are
> seeing the repository not simply as a database containing individual
> models, but as an dataset in and of itself. We need to work on this -
> I'm told Mike mentioned it at the Auckland CellML meeting last week.
>
> Hope you found this of some interest,
> James
>
> Abhishek Tiwari wrote:
>> As growing no of models in CellML repository I feel that CellML
>> Repository can be made more user friendly by implementing one or more
>> suggestions mentioned below-
>> 1. A Tree like browsing mechanism in which end child node of tree will
>> have model. For that we need to have classification system for the
>> models based on Physiological system or better say using MeSH (Medical
>> Subject Headings,something like
>> http://www.nlm.nih.gov/cgi/mesh/2008/MB_cgi). Models can be in multiple
>> classes and basically model metadata can have details for the
>> classification purpose like keywords. Domain ontologies can be used to
>> give better description for tree like representation.
>> 2. For a given model at the end of overview similar models (basic or
>> improved models or more complex models) can be reported (Similar to
>> model variants).
>> 3. Web services to search and download CellMl models for non ABI tools (
>> using CellMl API or Web API for CellMl).
>>
>> I don't know if efforts to handle above is under development or already
>> exists so I thought to write you all.
>>
>> abhishek
>>
>> _______________________________________________
>> cellml-discussion mailing list
>> cellml-discussion at cellml.org
>> http://www.cellml.org/mailman/listinfo/cellml-discussion
>>
>
>


--------------------------------------------------------------------------------


> _______________________________________________
> cellml-discussion mailing list
> cellml-discussion at cellml.org
> http://www.cellml.org/mailman/listinfo/cellml-discussion
>



Archive powered by MHonArc 2.6.18.

Top of page