Pedro Mendes wrote:
> > A -molecular- system should never be too complex to be described
> > successfully in elementary physico-chemical terms (mass action law,
> > diffusion, ...).
>> Petr, I think you could actually give us some arguments contrary to that
> view :) Very often kinetic data is not good enough to discriminate between
> models, I'm sure you agree with this. Plus, there is always *another*
> mechanism that will fit the data. Even the (in)famous Occam's razor does
> not say that the simplest mechanism is the TRUE one, just that if we
> cannot distinguish them WE ARE BETTER OFF considering the simplest.
Pedro,
thank you for your gentle explanations, and for the list of references.
I asked my question in the group, and not privately, hoping there might
be at least some people occasionally reading 'bionet' for general
education. Clearly, I am stray mechanistic bichemist here, but
unfortunately I don't even dare to consider the formation of
'bionet.kinetics-and-mechanism' group :) The traffic probably would be
dismally low.
Now on the issue of model discrimination and Occam's Razor, I am happy
to report that one of the true experts in the field of statistics (my
opinion), Bill Jefferys of the University of Texas, has agreed to embark
on a project with me just as soon as he finishes his department
chairmanship. Next month, we will start developing a general approach
to model discrimination in enzyme kinetics, based on the Bayesian
conception of probability.
For a nicely readable treatment of a related subject (although a rather
simple one, compared to what's coming our way) see
"Ockham's razor and Bayesian analysis."
Jefferys, W. and Berger, J. (1992)
`American Scientist', 80, 64--72.
Stay tuned...
-Petr
_____________________________________________________________
Petr Kuzmic Ph.D. * BioKin Ltd. * Madison, WI 53708-8336, USA
pkuzmic at biokin.com * http://www.biokin.com * 608.256.1269 fax
