I think this is a classic SBML discussion. Someone wants to do
something and rather than addressing how SBML might address it
and then discussing consequences, the discussion has turned to
whether his science is right. Other examples:
1. Should fractional stoichiometry be allowed? Luckily, the
flux guys finally won out.
2. 2d vs. 3d volumes.
3. Some other issue I've forgotten that had the stochastic
guys stuck for a year. It might be item 2.
It is true that one can't allow every oddball corner-case wild-eyed
scheme to take apart the fabric of SBML; however, that does not seem
to be the situation here. And I deeply believe that SBML is not the
venue for saying what is good science or not. Take him apart at the
seams at a conference or in response to a paper. But I think the
purpose of this venue would be to ask:
1. What exactly is being requested and can it be done already somehow?
Here, stoich for modifiers.
2. If it is impossible now, what possible ways could it be achieved?
Forget good/bad here, just list the ways.
3. Now, evaluate good/bad with regard to whether the possible
solutions harms someone elses work or pushes the system
into complexity risk.
If we carried stoich in association with "modifiers" would that
*harm* someones present or anticipated work? It may offend you
greatly, but will it harm you?
On Tue, Apr 26, 2005 at 10:25:26PM -0500, Howard Salis wrote:
> Without going into all of the theory, let me just say that there exists
> a probabilistic version of the quasi steady state approximation. Rao &
> Arkin have written a paper on one version of it as applied to the Master
> equation. And, like I mentioned in a previous post, the XML parsed
> MathML syntax (or something similar) that represents equations in SBML
> is not (at least for my purposes) an adequate way of storing the rate
> laws. Again, I like the idea of using a mapping of unique integers to
> their corresponding rate law forms, but I'd be happy with anything that
> works well.
> -Howard Salis
> Pedro Mendes wrote:
> >On Tuesday 26 April 2005 08:36 pm, Howard Salis wrote:
> >>Like the previous email, I was referring to what happens when you make
> >>the QSS approximation and write down the Michaelis Menten-like rate law.
> >>Like you said, I is then a modifier, and the rate law depends on its
> >>stoichiometry (among other things). If that information is not encoded
> >>in the SBML file, then a simulator using a continuous time MC method
> >>will not be able to simulate the system as intended by the modeler.
> >the information is encoded in the SBML file exactly as a rate law. It will
> >contain one or more terms for I with several constants. The stoichiometry
> >of I in the elementary reactions is reflected in these terms. However in
> >this overall reaction I has stoichiometry zero.
> >I believe from this last explanation, your problem is not having the
> >correct rate equation that reflects the behavior of the system with very
> >small numbers of molecules. In these conditions the steady-state
> >approximation (or the fast equilibrium one) is not appropriate and so you
> >will have to derive an overall rate equation in some other way, if at all
> >possible. I would suggest that you include all of the elementary steps. In
> >that case, my previous point holds (I would not be a modifier). If you
> >want an overall rate law that is appropriate for the conditions you
> >described then the solution does not pass by assigning stoichiometric
> >coefficients to modifiers. This is not a problem of SBML.
> >Hope that this helps with the solution.
Ed Frank 630-252-4548 (-5986, fax)
Division of Math & Computer Science Argonne National Laboratory