David Fell wrote:
>Paul Schlosser wrote:
>>If I can build a simulation model and put in everything I know about the
>>enzyme kinetics, ....
>I think Paul is about right in the first part, though reason 2 has not been
>a very significant factor for some time now.
That was kind of my point, though I was trying to be subtle.
>However, when he says 'If .....' of course his argument is pretty
>unassailable provided his conditions are met. But:
>>1. Michael Savageau pointed out (in 1976 I think) that the enzyme glutamine
>synthase is known to interact with about 15 metabolites ....
Well no model is perfect. But if the concentrations of those 13 of the 15
metabolites are not included in the matrix method, is it any better than a
simulation that also ignores them/assumes them to be 'in excess'?
>2. There are too many examples of perfectly respectable in vitro kinetic
>data not leading to sensible answers when used to calculate in vivo
>behaviour ....
Which is why simulation models are mostly calibrated against/fit to
in vivo data.
>I don't want to appear pessimistic. I believe we should use available
>biochemical knowledge to construct simulations exactly ....
You, and Athel, seem to be arguing that it must be an all-or-nothing choice.
If you allow me the same simplifying assumptions that are implicit in a
matrix approach, and the same original data/information, I still say the two
can produce equivalent results to specific questions.
To quote a colleague here, "A model doesn't have to be perfect to be useful,
it just has to be better than what has come before."
I think the key is in being clear on the intended use. This comes back to
my response to Pedro: there are things you can learn from the matrix approach
that are independent of kinetics, for example, that would be difficulat at
best to learn from simulations. But obtaing the flux CC for a specific step
in a specific pathway in a specific millieu (sp?) is not one of them. My
work now focuses mainly on the specifics.
-Paul
