[ IRIDIA
]
Biological Networks - From Structure to Dynamical Diversity
|
|
Intro
An important part of IRIDIA researches is
focused on the analysis of generic properties
of biological networks, i.e. properties that are
shared by networks and that can be found at
different levels of organization.
Our main aim is a qualitative and
quantitative characterization of the networks
and the study of how differences in the network
structure (i.e. its connectivity) influence
the repertoire of dynamical behaviours of the
entire network, its stability and adjustment
to environmental variations
Compartmentalization and networking are
two primary design principles in biological
systems. While compartmentalization has
been well studied for more than a decade
leading to one of the cornerstones
of modern biology and
medicine, networks have only
recently attracted the attention
of researchers. The network
organization is almost ubiquituous
in biological systems and
can be found on all structural
levels, from physiological,
immune and neural networks
to social organization of
insects, birds, mammals and
man up to the all encompassing
ecological networks at the
highest hierarchical level.
Despite many theoretical and
experimental efforts, the current
knowledge of networks
remains rather limited. One
possible explanation, besides
the complexity of the subject,
lies in the fact that researchers
have tended to emphasize
those aspects of behaviours
of networks that distinguish
them - instead of considering them from a
single perspective and thus identifying their
common features.
Only recently has it become clear that there
are many common properties, shared by various
types of networks.
whatís a biological network
A network can be viewed as a set of largely
identical subunits that interact, i.e.
communicate, with each other. Once the collection
of these sub-units has been identified, three
important properties that govern the behaviour
of a network can be distinguished: (a)
the connectivity of the network that determines
which subunits interact with which other
subunits, (b) the strength and nature
of these interactions, and (c) the total size
of the network. The acknowledgement that all
networks can be seen from this common
perspective constitutes a considerable reduction
of the problem's complexity. When the
problem is faced in this way, it is possible
to concentrate on precisely these parameters
and the types of sub-units under investigation,
and define further the classes of similarity
into which the diverse types of networks
can be placed.
This is the conceptual basis
that enables us to discover and examine common
rules governing different biological
networks. If this can be realized, we can
expect to develop a theoretical framework
for biological networks which plays the same
role as the modern theory of phase transitions
does for physical networks.
What to study
In our collaborative study we particularly
approach the following set of questions:
-
The dynamics of a biological network has
to be seen in the context of the relevant
biological environment. In this framework,
the dynamical behavioural patterns
become functional. Thus, biological networks
accomplish certain tasks and solve
problems. The interaction between the
elements of a network bears certain
characteristics of information transfer, i.e.
communication. Complex coherent functional
patterns emerge once a critical
communication level is reached. What are
the structural limitations on the possible
pectrum of dynamical activity patterns
displayed by a network?
- The dynamical properties of a network
sometimes change abruptly as one of the
network parameters is gradually varied.
These abrupt changes can be viewed as
phase transitions or bifurcations. What
principal kinds of phase transitions and
bifurcations are involved with biological
networks?
- An important further question in this
context is what structural features make a network
stable and allow it to perform its
function within the larger framework of
the organism in which it is located or
within the environment, since the network
must be stable and functional with and
without external perturbations.
- On a longer time scale, the structure of
the network changes due to evolutionary
processes. The latter are responsible for
the plasticity of the immune system or an
ecological system. Evolution of networks
allows these systems to adapt to varying
environmental conditions. Some of the
networks we intend to study have been
chosen because they can evolve. Indeed,
the immune system of a child is different
from that of an adult. Moreover, when
such systems interact with other evolving
networks, they show co-evolution, i.e.
parallel evolution in which mutual
constraints between the evolving nets
are reduced. In the presence of retroviruses,
e.g. there will be co-evolution between
the immune system and the retrovirus population.
Within a short period, the immune system has
to reshape in order to compete with and
eventually suppress the retrovirus population.
What laws govern evolution and co-evolution
of biological networks?
- At the same time, despite this rapid evolution,
the immune system must maintain its functional
integrity and interact with the genome, the enzyme
system and the hormone system, all three of which
do not evolve during the life time of an individual.
There is thus interaction between evolving and
non-evolving networks. Similar remarks apply to the
ecological systems and systems that are located
higher in the hierarchy, e.g. for social systems
and insect societies. What are the structural
requirements that enable evolving and co-evolving
networks to maintain reliable operational
conditions?
How to connect computer simulations with experimental data
There are two ways to obtain data for the
behaviour of biological networks:
computer simulations and experimental work.
Computer simulations are necessary in order
to guarantee "clean" and reproducible conditions
in which it is possible to investigate problems
that cannot be solved in real laboratory or
field experiments.
They are especially suited to uncover generic
network properties such as scaling and phase
transitions in such a way as to fulfill demanding
scientific requirements such as reproducibility.
Our work is basically theoretical and our task is
to devise a mathematical theory. However, close
links with experimental groups are maintained.
Our research aims at unveiling these parallels
and at developing tools and concepts to foster
a generalized theoretical framework for these systems.
Hence, at present the project represents purely fundamental research.
It is highly desirable to put the expected results to
work in an application oriented framework someday.
However one has to distinguish between application in
general and involvement of industry in
particular. Several aspects of our research that are
oriented towards the elucidation of ecological networks
may find useful applications at the level of environmental management
and may help to define decision criteria for governmental
use and industrial and societal implementaion.
There will certainly be important issues at the level
of the biochemical networks that we plan to study
and that can find application in medical research
and the pharmaceutical industry.
We do expect that contacts with industry will become more successful,
after the goals we try to achieve in this research - i.e.
an increased understanding of complex networks that goes
along with the development of tools to manipumate these s
ystems - have been reached.
Researchers in IRIDIA have been working together for
many years as members of the Paris-Brussels group on
Theoretical Immunology in which scientists of the Ecole
Polytechnique in Paris (Varela, Calenbuhr), the Pasteur
Institute (Coutinho, Stewart, Carneiro) and the UniversitÈ
Libre de Bruxelles (Bersini, Calenbuhr, Detours) participate.
IRIDIA was basically involved in modeling complex dynamical
aspects of the immune system in relation to auto-immune
disease and other pathological disarrays as well as in
relation to early development of the immune system in young
individuals. The approaches and techniques used are
differential equations, qualititative study of complex systems,
bifurcation theory, computer simulation and cellular automata.
[ Hugues Bersini | Vera Calenbuhr | Vincent Detours ]
Selected references
- Detours, V., Bersini, H., Stewart, J. and F. Varela :
- Development of an Idiotypic Network in Shape Space.
in Journal of Theoretical Biology - 1994 - 170.
- Bersini, H. and V. Calenbuhr :
- Frustration Induced Chaos in a System of Coupled ODE's.
in Chaos, Soliton and Fractals - 1995 - August.
- Calenbuhr, V., Bersini, H., Stewart, J. and F. Varela. :
- Natural Tolerance in a Simple Immune Network
Accepted for Publication in Journal of Theoretical Biology. 1995
[ Main
| Projects
| People
| Meeting
| Library
]
Last updated Jan 15, 1996 |
Comments to webmaster