67

advantage of the method lies in the fact that it is possible to model processes without pre­

cise data on the speed (“kinetics”). If, on the other hand, one wants to model a dynamic

process, in particular a signal cascade, in more detail, one must determine these data about

the velocity. This is done by methods of time series analysis: If one has measured the pro­

cess (for example the phosphorylation of a kinase that transmits a signal in the cell) for

five or more time points, there is enough data to estimate how fast this process proceeds.

It is therefore possible to describe the speed (kinetics) precisely in mathematical terms

using a parameter (in the example: the speed). There are a number of bioinformatics tools

for estimating parameters. Easy to learn and good to use for this parameter estimation is

the software Potters Wheel (https://www.potterswheel.de/Pages/; Maiwald and Timmer

2008) and its successor Data2Dynamics (Steiert et al., 2019).

This software can also be used to investigate which parameters need to be accurately

estimated and which do not (sensitivity analysis). It also allows to see which of the param­

eters can be well estimated from the data (identifiability analysis) and which cannot (either

because the data are not sufficient or because the network is wired in such a way that, for

example, the parameter always depends on another one that cannot be estimated either or

because the parameter is simply not determined by this data at all).

Conclusion

• Systems biology modelling of signalling cascades and protein networks allows deeper

insights into the function of the proteins involved and thus helps to understand the

causes of diseases, to better describe infection processes and immune responses, or to

elucidate complex processes in biology, such as cell differentiation and neurobiology.

Stronger mathematical models describe signalling networks precisely in terms of

changes over time and their speed using differential equations. This explains the pro­

cess exactly, but additional time is needed, e.g. with the determination of the velocities

(kinetics, data driven modeling, time series analysis).

• Boolean models only require information about which proteins are involved in the net­

work and which protein interacts with which other proteins and how (activating or

inhibiting). Therefore, they are well suited for an introduction. If you want to reproduce

one of the presented examples yourself, it is easy (use the same components and links

and software). However, if you want to create your own new model, many cycles are

necessary, because you have to check again and again in simulations based on the

Boolean model (e.g. with SQUAD or Jimena) whether the behaviour in the computer

model also matches the outcome actually observed in the experiment, at least qualita­

tively, and thus adapt the computer model to the data step by step.

• Conversely, the model then allows to describe all situations that have not yet been mea­

sured or reproduced in the experiment. In particular, the effect of drugs and their com­

binations, the activity of all proteins involved, the effect of signals, mutations or even

immune substances (e.g. cytokines). Systems biology modeling can be described as the

central, current field of bioinformatics. It is also called network analysis, dynamic mod­

elling or interactomics in order to emphasize these aspects more strongly. ◄

5.2  Generalization: How to Build a Systems Biology Model?