formulation of new hypotheses about the effect of specific internal
or external perturbations in a system [40]. Using a systems biology
approach (Fig.2), the iterative cycle of data-driven modeling and
model-driven experimentation refines formulated hypotheses until
they are validated.
“Models” are an abstract representation of reality which pro-
vide a reliable sense (i.e., understanding the behavior) of the origi-
nal system depending on available information and the purpose of
modeling [6, 40, 41]. “Modeling” is the process of creation and
usage of a model [41]. Mathematical models describe the reality
(i.e., processes in a cell) in terms of functions or equations which
contain variables and parameters:f(X 1 ,...,XN;k 1 ,...,kN), wherefis
the function that evaluates, e.g., the temporal behavior of a system
depending on the variablesX 1 ,...,XNand parametersk 1 ,...,kN.
Variables are the quantity of interest in model analysis which typi-
cally change over time, e.g., concentration of protein in a cell.
Parameters are quantities which are fixed for a given computational
experiment to characterize specific quantitative behavior of a
model. Parameter values are typically characterized from the litera-
ture, databases, for example BioModels and SABIO-RK, and can be
estimated from experimental data by calibrating the model to reca-
pitulate the real biological process [15, 42]. After calibrating the
model with certain experimental data, it can perform a large set of
repetitive in silico experiments for many different conditions that
may be quite time-consuming and expensive with wet-lab experi-
ments. Models can be created at different levels of abstraction,
ranging from coarse grained qualitative models of (large) subcellu-
lar processes to a detailed quantitative model of a (small) functional
module (Fig.7).2.4.1 ODE-Based
Modeling
If the relevant components in a network are largely known and
sufficient quantitative data available, for a small-scale network,
ODE-based models are widely used to analyze the functional role
of nonlinear biochemical networks [43–45]. In such models the
reactions are represented by a set of differential equations (Fig.7)
describing change in quantity of reactants to products and vice
versa, in case of reversible reactions. Based on reaction rate and
kinetic parameters such models usually yield high-quality predic-
tions of the system’s dynamics with quantitative information about
molecular concentrations [46]. These models, however, require
accurate kinetic parameters which is often infeasible for large net-
works; therefore, the ODE-based model of large biochemical sys-
tems is very difficult if not possible. In such cases the Boolean/
logic-based model is a suitable option [47, 48].2.4.2 Logic-Based
Modeling
Logic-based modeling is a popular approach to describe the quali-
tative temporal behavior of a large system of interactions where
experimental data are frequently sparse (not all can be measured,Integrative Workflow for Predicting Disease Signatures 257