Dear Sammys and Marys
I have been following the discussion of Paul's 'direct simulation'
thread with interest. As is often the case, a well-directed question,
splits the world into two - here, the factuals and the conceptuals
(yes, I know I oversimplify, but what the hell, I like to generalise).
I dreamt last night a parable concerning two friends, Sammy Simulator
and Mary M.C.A. Modeller (Mary was blessed with middle names): Sammy
built a beautiful detailed wooden model of a 2nd World War RAF
aircraft with everything down to the last T; he could fiddle with
everything, see what happened to the wingflaps when he changed the
position of the joystick, and much more. He went on to become the
world's expert on this specific plane and was very famous. It didn't
matter much that his model never flew. Mary, on the other hand, was
passionately interested in flying. She started off building simple
paper planes to discover the basics of aerodynamics. She asked many
'What if'-questions: what if the plane had no tail?, how does
wingshape affect the performance?, etc. Her models were a far cry from
the real thing, in fact some of them looked like nothing we know, but
they flew beautifully around her garden (she never got as far as
building a real big aircraft). By playing around with numerous
variations on many themes she discovered the general principles of
aerodynamics and in the end got to write the definitive textbook on
the subject. Fortunately she also became famous.
When I woke up I realised that we need both the Sammy's and the
Mary's. Nevertheless, we should also realise that they need different
tools, or, if they use the same tool, each will use it for a different
purpose.
Now to get more specific (and rude?): I am 100% a Mary, have always
been, will always be. By nature, I am a bit skeptical about Sammys,
(1) because I think they too often fall in love with their beautiful,
detailed simulations, and (2) I've never seen anything really new
coming from such simulations (perhaps I shouldn't be expecting this,
but my Maryness overwhelms me). Nevertheless, many of my best friends
are Sammys, so it doesn't bother me too much.
Now, to satisfy my Mary-nature I use both MCA and computer simulation,
but I've never built a detailed Schlosser-type simulation of anything,
although I do take great pains to make my 'paper-plane' models
realistic in the general chemical and biological sense (for example, I
never define a rate equation as irreversible, except perhaps for a
sink to the outside --- because we have been forced in the past to
violate this principle Athel and I found it necessary to develop a
reversible Hill-equation - CABIOS real soon now ).
As for MCA, to claim that it is a subset of a detailed model is to
underestimate its uses severely. A sensitivity analysis of a model is
but a small, albeit important, part of MCA. What I find so exciting
about MCA is the power it affords me for conceptual and general
reasoning about metabolic behaviour. As Athel and I stated in our
contribution to the Louvain-la-Neuve BTK meeting, there are
unfortunately too few examples of this use of MCA in the literature,
most of which concentrates on the measurement of control coefficients
(and, of course, the theoretical development of the framework). Our
paper aimed to give one such example: without any modelling whatsoever
we could show with MCA that when a metabolite regulates its synthesis
by direct product inhibition (as serine does in mammals), the only
type of inhibition that would be effective is uncompetitive inhibition
(the type with which serine inhibits phosphoserine phosphatase).
So, my fellow Sammys, I'll continue to watch your efforts with great
interest and will gladly learn from them. But, in the meantime, I
merrily mary along.
Jannie
