Re: blog: semantic dissonance in uniprot

On Mar 30, 2009, at 8:15 AM, Oliver Ruebenacker wrote:

>     Hello Pat, All,
>
> On Sun, Mar 29, 2009 at 11:23 PM, Pat Hayes <phayes@ihmc.us> wrote:
>> On Mar 29, 2009, at 10:48 AM, Oliver Ruebenacker wrote:
>>>  Perhaps the question should read: What would you advice to some one
>>> who wants to build an ontology to describe pathways for Systems
>>> Biology purposes?
>>
>> I really have no advice to give, as I know virtually nothing about  
>> systems
>> biology. My remarks were based on your raising the topic of  
>> statistical
>> ensembles, which is enough to make me want to go and do something  
>> else, I'm
>> afraid.
>
>  There is no reason to be scared of statistical ensembles!

Pat Hayes isn't scared of them, but PatHayes the ontologist is very  
leery of trying to describe them using ontologies written in any logic- 
based language. We don't (yet, AFAIK) have a satisfactory ontology for  
things like variances and expectations and probabilities. None of this  
has been adequately formalized yet, and this whole area is a logical  
minefield: for example, reasoning about approximate values is a well- 
known problem area, rife with potential "paradoxes" like the heap  
argument.

>  Forget for the moment about Systems Biology and think of Science in  
> general.
>
>  A scientist applies a method to obtain results. Talking about the
> method means talking about things particular to the discipline and the
> tools used (e.g "On March 14, I placed the soccer ball on my table and
> held my yellow ruler next to it.").
>
>  The results can also be expressed in this language (e.g. "My yellow
> ruler read 27.3 inches when held next to the soccer ball on my
> table"), which is useful for those interested in the method, but it is
> not required for those only interested in the results. In fact, if we
> want to collect, combine and compare results obtained by different
> methods, details particular to methods are usually counter-productive
> (ruler? measuring tape? caliper? laser? red? yellow? table? chair?).
> The consumer of the results is typically interested in a statement
> such as "the diameter of a soccer ball is 27.4 inches plus-minus 0.3
> inches". An expectation value and a variance.
>
>  But what does the "diameter of 27.4 inches plus-minus 0.3 inches" in
> above statement inhere in? Certainly not one particular soccer ball.
> How about the set of all the soccer balls in the world? But why would
> we be interested in all the soccer balls in the world?

Because you want to make _general_ statements about soccer balls. That  
"general" is another way of saying "All ... in the world". And general  
statements are what ontologies largely consist of.

> What if in some
> strange place, people produce soccer balls of unusual sizes, would
> that change the result for us?

If they really were soccer balls, yes. But you could always invent a  
new category of "strange balls similar to soccer balls".

>
>  What we really mean when we say that a soccer ball has a diameter of
> such and such, is that we imply that there are standard ways to obtain
> soccer balls (e.g. going to the next sports store and buying one, or
> taking one from the equipment room of the next gym, or participating
> in a soccer match), and that these ways are equivalent in the expected
> results.

Hmm. That is not what I mean when I say that. What I mean is, all  
soccer balls have a diameter of such and such. I don't accept the  
premis that all statements are reducible to statements about  
experimental scenarios (even in science, although I will grant you  
that this is more plausible for quantum-theoretical physics, if one  
accepts the Copenhagen interpretation. I prefer the transactional  
interpretation, myself.)

> So there will be a range of possible scenarios (e.g. a soccer
> ball of 27.2 inches, a soccer ball of 27.342 inches, a soccer ball of
> 9*pi inches, etc.) and each outcome comes with a probability. These
> probabilities are usually not all the same, and since the number of
> possible scenarios is typically infinite, it makes no sense to assume
> they are. The probability distribution over the set of possible
> outcomes is a ensemble.

Yes, I know. I also know that I have absolutely no idea how to  
adequately formalize this so that its salient properties can be  
deduced from the formalization. I also, BTW, have strong doubts that  
this is the best framework within which to attempt to formalize  
biological information, but no doubt we would have to agree to  
disagree on that point.

Pat

>
>  Enter Systems Biology. Systems Biology is a consumer of results from
> many disciplines. Its objective is to put together results obtained by
> a wide range of experimental, theoretical and computational methods in
> Biological Physics, Molecular Biology, Physical Chemistry,
> Biochemistry, Computer Science and Applied Math. Some of these methods
> involve tracking single molecules, of definite sets of molecules. Some
> do not. so Systems Biology needs a language to talk about these
> results without referring to artefacts of methods used, which is the
> language of ensembles.
>
>     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
>
>

------------------------------------------------------------
IHMC                                     (850)434 8903 or (650)494 3973
40 South Alcaniz St.           (850)202 4416   office
Pensacola                            (850)202 4440   fax
FL 32502                              (850)291 0667   mobile
phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes

Received on Monday, 30 March 2009 15:49:49 UTC