Information Technology Reference
In-Depth Information
explain the way in which these networks were constituted in time, again biology is
clearly affected. When a idiotypic network growing in a real shape space spontaneous
separates this space in a “immune” and a “tolerant” zone, although the concentration
update mechanism is everywhere the same [3], again this result is far from expected
and highlight in new ways the fundamental self/non-self distinction of immunology.
And whatever experiment has surprising outcome, scientists will show their face with
great expectation. When Perelson et al [26] explain by a simple mathematical model
the antagonistic population dynamics of CD4 T cells and HIV virus and qualitatively
replicate the long life time of T cell despite the huge presence of virus, again the im-
pact is important. Although still to be construed in a qualitative way, the reproduction
of these phenomena by software means help to unveil the underlying mechanisms
giving rise to them.
Computer language, although very rigorous, offers more flexibility than any ma-
thematical language. The computer can replay certain scenarios of biological evolu-
tion which have taken place over millions of years endlessly, without the programmer
having to resort to gnawing at the mouse. This allows the scientist to test several hy-
potheses, retaining only that one which resembles the current situation most closely.
The programmer creates new worlds, worlds which evolve on their own and he can,
as necessary, select those which are worth allowing to evolve somewhat. The compu-
ter suggests a result and the scientist adapts to it, looking to understand the result and
ensuring that it is not a simple artefact linked to the intrinsic limitations of the method
of inputting and processing digital information. At last, the “Grail” to reach for any
scientific modelling attempt remains the quantitative prediction, a prediction so accu-
rate that a measuring device will be able to validate the modelling by comparing what
it measures with the model prediction. Several theoretical immunologists [26, 11, 12]
force the way to go beyond qualitative descriptions and to quantify the immune sys-
tem behaviour through mathematical and computer simulation approaches. As Rob
De Boer [11] claims: “
Theoretical immunology is maturing into a discipline where
modelling helps to interpret experimental data, to resolve controversies, and - most
importantly - to suggest novel experiments allowing for more conclusive and more
quantitative interpretations
”. Nevertheless, all other sorts of modelling whatsoever
qualitative, on the road to the ideal most predictive one and for reasons mentioned
above, like “pedagogy“ or the testing of “emergent phenomena”, are equally worth
the effort.
Since there is no reason why immunologists should be surprised or disagree with
these previous arguments, what would they gain in collaborating with researchers in
computer science well decided at contributing to this modelling enterprise? I see one
strong reason that I will expand below. The immune system is a very complex one,
full of chemical actors interacting in complicated ways. These last twenty years com-
puter scientists have been well trained for software simulation of complex systems by
learning and practising the “Object Oriented (OO)” tricks, tricks that biologists (natu-
ral scientists in general) still seem to be hesitant (mainly because not educated to) to
acquire and master. The OO software are simultaneously easy to read and to under-
stand (even for non programmers), simple to build, easy to modularize, to maintain
and to adapt. New software tools entirely rely on Object-Orientation (OO), essentially
OO programming languages (C++, java, .Net, Python), UML and Design Patterns.
From its origins, OO computation has simplified the programming of complex reality
