65

dimerization inhibitors against Erk dimerization – both ways to prevent heart failure at the

molecular level. Furthermore, the bioinformatic model also clarifies the downstream

targets (i.e. target proteins) of heart failure, which can also be pharmacologically influ­

enced to prevent or favourably influence heart failure.

Alternatives to Semiquantitative Modeling

If further data is known, especially about the velocity and stimulus strength in the signal

chain, the data-driven modeling can be taken even further and the exact velocities, affini­

ties and chemical equilibria can be calculated more precisely. With this, enough informa­

tion is then available to represent this process with exact equations, so-called differential

equations, which thus have the change of a quantity on the left side and describe this

change on the right side via the quantity itself and further determining factors. If I know

all the influencing factors, I know the constants and kinetic properties of the signal cascade

(in mathematical terms, this is called the “parameters” of the differential equation), and I

can then use them to model the system accurately and precisely. An example is the inhibi­

tory cAMP and cGMP signalling pathways in the platelet, which thus dampen the platelet

in its activation. Here we had enough information from experiments that we patiently

repeated over and over again for several years to set up such a model (Wangorsch et al.

2011). This area of accurate modeling is also being pursued by many systems biology

groups. A simple approach to set up such models oneself is the software PLAS (Power law

analysis and simulation; https://enzymology.fc.ul.pt/software/plas/), which also intro­

duces one to all the steps for this more accurate simulation via tutorials. However, as a

beginner you have to make many decisions about the parameters. But if there are too many

“free” parameters, one runs very easily the risk that the system is described incorrectly,

because one can always choose the free parameters in the equations in such a way that the

system seems to fit the little available data, but then very easily misses the mark with new

experiments or data. This is easily prevented in the semiquantitative models. This is

because these are coarser, but have fewer free parameters and therefore are not as quick to

be wrong in their predictions as the much more accurate quantitative models. Finally, it is

worth mentioning that one can also stop at step 1 and also just examine the structure of the

model in detail. This works for signaling cascades as well as for metabolism. For the latter,

glycolysis or the citric acid cycle, for example, are very illustrative textbook examples,

which are followed by further insights from, for example, the linear metabolic pathway of

glycolysis and from the cyclic pathway of the citric acid cycle for metabolism. An over­

view of these different systems biology methods and approaches is provided in the English

textbook by Klipp et al. (2016).

Finally, we have collected an introductory selection of our own work on Boolean

models and semiquantitative modeling based on them (see below), which should give an

overview of the basics, but also various examples of applications, and help the interested

reader to continue learning.

5.2  Generalization: How to Build a Systems Biology Model?