>Petr Kuzmic <pkuzmic at biokin.com> wrote in message
>news:399F568A.7D2A0EED at biokin.com...>>>> I would like to ask if anybody remembers the work of David Garfinkel in
>> computer simulation of metabolic pathways in the 1960's. This, to me
>> anyway, was a very serious attempt at applying computational techniques
>> to biochemical metabolic systems.
>>I'm sure several people in this newsgroup do. I have a good collection of
>his
>papers that span early 60s until the early 90s. But even further back was
>the
>work of Chance (I presume Garfinkel was one of his students) and Hess. From
>my archeological expeditions to the library I have concluded that the first
>paper
>on computer simulation in biochemistry was back in 1943 by Chance where
>he used a mechanical differential analyzer to simulate time courses of
>peroxidase.
Of course I do. I was part of the group from 1973 to 1984. Yes David
did start out with Britton Chance, and initially they did use
differential analyzers and then analog computers to model time
courses. However, it was too cumbersome and they abandoned that
effort.
> > It failed, more or less, because the *numerical methods* were not mature
>> enough then. I wonder what other contributors to this forum think about
>> the significance of Garfinkel's work.
>>I'm not sure it failed only because of the numerical methods. In fact,
>Garfinkel's
>group helped highlight the problem of stiffness and eventually used stiff
>integrators. I think the biggest problem that Garfinkel's work encountered
>(and other groups later on) is one that has not yet been solved completely:
>they were integrating parameters studied in vitro (and in isolation) in a
>system.
I dispute the contention that the effort failed. It's true that
advances in computer hardware, mathematical techniques (such as the
Gear-Hindmarsh integrator), and software were necessary before such
modeling could be done routinely. By the time I left Garfinkel, we
were readily building very large, detailed models of metabolism; see
Annals of Biomedical Engineering, 11, 361-384, 511-531, 533-549
(1983). The paucity of people with training and experience in
mathematics as well as biochemistry/physiology caused this field of
research to be regarded as a mere curiosity and not as something
others could use. With today's much more capable hardware and
software and more scientists than ever being comfortable with
computation, perhaps interest in large-scale biochemical systems
could be revived.
>It turns out that several of these parameters are not applicable to the
>system
>because of fundamental differences between cells and test tubes
>(concentrations,
>organization, etc.). I think that simulations of really large biochemical
>networks
>will be much more accurate when we do the parameter estimation in the actual
>real system.
I came across this problem when I was constructing my model of
pyruvate metabolism (Am J Physiol, 1979). Substrate channeling in
multi-enzyme complexes had to be included explicitly in the
equations. My work on a fatty acid metabolism model had to consider
structural factors. Some of the enzymes (e.g. fatty acyl CoA
synthetase) lost activity when they were purified from the membranes
in which they were embedded. But these problems are not
insurmountable, and such effects were represented in the model.
Today, I am studying metabolic activation of xenobiotic compounds to
carcinogenic intermediates. Here too, enzyme localization,
membrane-bound protein complexes, and substrate channeling have
pronounced effects on the carcinogenicity of an exotic chemical.
--
Michael C. Kohn
Laboratory of Computational Biology and Risk Analysis
National Institute of Environmental Health Sciences
P.O. Box 12233, Mail Drop A3-06
Research Triangle Park, NC 27709-2233
919-541-4929 (voice)
919-541-1470 (Fax)
E-mail: kohn at valiant.niehs.nih.gov
Web site: http://valiant.niehs.nih.gov
---
