Information Technology Reference
In-Depth Information
indicates that the topology of regulatory networks has a significant role in
restricting their dynamical behavior.
Boolean modeling approaches rely on the network topology and provide a
backbone to integrate different studies. Network topology can be improved by
perturbation studies and of course by additional data. In the current state of
knowledge in many systems, even the published data cannot be easily wired
due to the lack of causal information (refer to Fig. 4.5). Thus the network
assembly is already a step forward in defining a system and interpreting the
experimental data. Often there are debates in interpreting the model predictions.
The dynamical output of the network is a collective behavior of all the
components. It is impossible to use experimental outcomes such as “bacteria
deficient in the type III secretion system are cleared faster by the immune
system” as inputs to the model. Instead, specific information on the function of
the type III secretion system can be used to infer interactions that are then used
during the construction of the network. While there is no guarantee that the
dynamic behavior of the network will reflect the original outcome, the correctly
predicted dynamic behavior of the network serves as a validation of the wiring.
The second important characteristic of the Boolean approach is its usability in
analyzing large systems. The Boolean simplification into two states is often
unfavorably compared to continuous approaches. However, the latter approach is
only practical for small subsystems. Though properly chosen subsystem models
have been successful, they are inadequate to study the emergence of holistic
properties. Instead, it is often advantageous to employ a system-level network-
based modeling approach. The resulting network models not only provide
holistic insight but also predict the outcome of subsystems. The input to the
network is from completely independent studies and the model is verified by
comparing its dynamic (time-elapsed) behavior with experimental time-course
observations.
The Boolean approach helps us understand the dynamics of the system by
giving an activation pattern where the sequence of dynamic events can be
compared between simulated and experimental data. The reproduction of such a
sequence indicates the components important in various stages of the dynamics.
Some of the Boolean approaches also give insight about the timescale of the
processes (e.g. asynchronous Boolean models) and the activation thresholds (e.g.
threshold models) of the nodes. In a sense we can view them as coarse-graining
of the differential equation models that keeps the same steady states.
As we have illustrated in this chapter, Boolean modeling can successfully
describe a variety of networks from the molecular to the physiological level. This
does not mean that all networks are suitably described by Boolean models. For
