||Many reverse engineering algorithms have been presented in the literature to infer gene regulatory networks. Objective testing of these algorithms can be performed on synthetic data, generated from gene networks of known topology. Different models are available at this purpose, incorporating different characteristics of real regulatory networks. Differential equation based models provide continuous data, but in general address simplistic regulatory logic, e.g. additive or multiplicative effects. On the opposite, Boolean network based models describe important aspects of gene regulation such as complex concurrent regulation mechanisms, but do not describe continuous changes in gene expression. To test reverse engineering methods minimizing the bias toward the chosen simulation, it is important to develop simulators that, by integrating different modeling approaches, are able to resemble the main features of real regulatory networks.
A simulation model is proposed to generate time series expression profiles by integrating differential equation with Boolean logic, so as to reproduce some important features of regulation in real biological systems, such as: the scale-free topology and the modular organization of the network; the continuous nature of generated data; the interaction among regulatory genes in controlling both production and degradation. The simulator combines three different models to describe network topology, regulatory rules and expression dynamics, respectively.
- Network topology is generated as a hierarchical structure in which modules are replicated at different level of network organization (as in fractals) and interconnected maintaining a scale-free distribution of connectivity degree.
- A regulation model describes interaction among regulators that activate or inhibit transcription based on a combination of different logic rules. As a result, a target value is defined for each gene at each time, with expression in the continuous domain.
- Finally, a dynamic model is used to generate continuous time series expression data by using differential equations, which describe how each gene tends to its target value in terms of gene-specific dynamic parameters influencing the rate of transcription and degradation.
The model is also able to generate both network topologies and time series data with characteristics similar to those of real biological systems, which makes it suitable to test reverse engineering algorithms.