Quantitative Science with OBO?

[Oliver Ruebenacker]

Hello, All,

We are trying to use the OBO approach to master a very complicated
subject: pathways. Recently, I realized that I don't even know how to
use OBO to do a simple example of quantitative science.

Suppose, I want to fly with my helicopter from Boston to either
Washington DC or New York City, whichever is closer.

Suppose, I don't know which one is closer. Tell me using OBO.

Take care
Oliver

--
Oliver Ruebenacker, Computational Cell Biologist
BioPAX Integration at Virtual Cell (http://vcell.org/biopax)
Center for Cell Analysis and Modeling
http://www.oliver.curiousworld.org

[Alan Ruttenberg]

Hello Oliver,

The OBO approach is for representation, and within that realm, some
level of computation. In an example we worked with recently, I
presented the use case from OBI of how one might represent the
measurement of a concentration.

https://obi.svn.sourceforge.net/svnroot/obi/trunk/src/examples/development/concentration-demoed-vancouver-2009.owl

OWL can not currently do any arithmetic. What we might expect to be
able to represent are things like positions of airports on the globe,
measured times for flights between cities, distances between cities
"as the crow flies".

Based on a) such representations, b) a query language which is able to
retrieve what we have represented and certain implications of it (we
have been using sparql on the inferred graph) and c) a language for
computation, offering operators such as comparison, subtraction, etc -
we would be able to answer a query such as the one you ask for.

-Alan

[Michel Dumontier]

Oliver,

  Ontology is descriptive. We can assert that boston and NYC and DC are cities. We can store the distances in the ontology along with their unit of measurement. We can categorize the distances as short or far, according to some definition. If you want to compare values, you must use a higher order logic like rules or a query language like SPARQL.

-=Michel=-

On Mon, Mar 30, 2009 at 9:45 PM, Oliver Ruebenacker <cur...@gmail.com> wrote:

--
Michel Dumontier
Assistant Professor of Bioinformatics
http://dumontierlab.com

[Oliver Ruebenacker]

Hello, Alan, Michel,

I think the most interesting part would be how you define in an OBO
way what the distance between two cities is.

Take care
Oliver

[Alan Ruttenberg]

On Mon, Mar 30, 2009 at 10:23 PM, Oliver Ruebenacker <cur...@gmail.com> wrote:
>
>     Hello, Alan, Michel,
>
>  I think the most interesting part would be how you define in an OBO
> way what the distance between two cities is.

The distance between two cities would be a relational quality.
http://www.ebi.ac.uk/ontology-lookup/browse.do?ontName=PATO&termId=PATO%3A0000040&termName=distance

There is an issue in that you need to say what kind of distance you
are defining (as the crow flies -
http://en.wikipedia.org/wiki/As_the_crow_flies), but this can be done
by making a subclass of the PATO class. You might want to coordinate
with the GAZ people: http://gensc.org/gc_wiki/index.php/GAZ_Project

To record the quantitative difference you might ask if there is a
practice within the phenotype community. Or you can use the pattern we
used for concentration which is also a relational quality, i.e. say
that the quality inheres in one city, towards another and that there
is a measurement about that quality that has some unit label (perhaps
http://www.ebi.ac.uk/ontology-lookup/init.do?termId=UO%3A0000008) and
some value.

Properties are currently defined in
http://purl.obofoundry.org/obo/iao.owl (as per the concentration
example), but will show up in RO at some point.

-Alan

[Michel Dumontier]

Not necessarily the "OBO way" (whatever that is), but here's a simple ontology to represent the info.

-=Michel=-

[Oliver Ruebenacker]

Hello Alan, All,

On Mon, Mar 30, 2009 at 10:42 PM, Alan Ruttenberg
<alanrut...@gmail.com> wrote:
> On Mon, Mar 30, 2009 at 10:23 PM, Oliver Ruebenacker <cur...@gmail.com> wrote:
> The distance between two cities would be a relational quality.
> http://www.ebi.ac.uk/ontology-lookup/browse.do?ontName=PATO&termId=PATO%3A0000040&termName=distance

I don't know what "extend of space between two objects or places" means.

> There is an issue in that you need to say what kind of distance you
> are defining (as the crow flies -
> http://en.wikipedia.org/wiki/As_the_crow_flies),

As the crow flies.

[Alan Ruttenberg]

On Mon, Mar 30, 2009 at 11:13 PM, Oliver Ruebenacker <cur...@gmail.com> wrote:
>
>     Hello Alan, All,
>
> On Mon, Mar 30, 2009 at 10:42 PM, Alan Ruttenberg
> <alanrut...@gmail.com> wrote:
>> On Mon, Mar 30, 2009 at 10:23 PM, Oliver Ruebenacker <cur...@gmail.com> wrote:
>> The distance between two cities would be a relational quality.
>> http://www.ebi.ac.uk/ontology-lookup/browse.do?ontName=PATO&termId=PATO%3A0000040&termName=distance
>
>  I don't know what "extend of space between two objects or places" means.

s/extend/extent/

It means "distance" ;-)

But there is a reasonable question as to what is exactly meant as the
distance between two cities. The appropriate approach would be to make
a subclass of distance that further specifies exactly what *you* mean.

If you don't find PATO's definition understandable, here's their tracker:

http://sourceforge.net/tracker/?group_id=76834&atid=595654

Please submit a comment.

-Alan

[Oliver Ruebenacker]

Hello Alan, All,

On Mon, Mar 30, 2009 at 11:28 PM, Alan Ruttenberg
<alanrut...@gmail.com> wrote:
> On Mon, Mar 30, 2009 at 11:13 PM, Oliver Ruebenacker <cur...@gmail.com> wrote:
> But there is a reasonable question as to what is exactly meant as the
> distance between two cities. The appropriate approach would be to make
> a subclass of distance that further specifies exactly what *you* mean.

You mean distance can mean whatever I mean it to mean? ;)

How do I know that the intended meaning of their "distance" includes
my "distance"?

> If you don't find PATO's definition understandable, here's their tracker:
>
> http://sourceforge.net/tracker/?group_id=76834&atid=595654

Good idea. How often do they typically release a new version?

Do you think I need to worry about how to define "Boston"?

[Alan Ruttenberg]

On Tue, Mar 31, 2009 at 11:42 AM, Oliver Ruebenacker <cur...@gmail.com> wrote:
>
>     Hello Alan, All,
>
> On Mon, Mar 30, 2009 at 11:28 PM, Alan Ruttenberg
> <alanrut...@gmail.com> wrote:
>> On Mon, Mar 30, 2009 at 11:13 PM, Oliver Ruebenacker <cur...@gmail.com> wrote:
>> But there is a reasonable question as to what is exactly meant as the
>> distance between two cities. The appropriate approach would be to make
>> a subclass of distance that further specifies exactly what *you* mean.
>
>  You mean distance can mean whatever I mean it to mean? ;)

If you make it a subclass of another class you have to make sure that
the definition of the superclass is also true of the subclass.

You can make "distance" mean whatever you want it to mean. You might
want the preferred term be something that won't be confused by others.

However, absent a choice on your part, words are ambiguous and it's
important, when using words to talk about classes, that you make clear
which class you mean.

http://www.golovchenko.org/cgi-bin/wnsearch?q=distance

>  How do I know that the intended meaning of their "distance" includes
> my "distance"?

If the documentation of the class doesn't make that clear enough,
submit a report. Talk to the developers. Document your meaning very
well so others don't have to feel the same way about your definition.

>
>> If you don't find PATO's definition understandable, here's their tracker:
>>
>> http://sourceforge.net/tracker/?group_id=76834&atid=595654
>
>  Good idea. How often do they typically release a new version?

Very often.

>  Do you think I need to worry about how to define "Boston"?

Document. Yes. Instances typically are not "defined".