Based on the linearization scheme presented previously, a linear state-space method can be applied to model gene networks.
In the past decade, various methods have been developed to model gene networks, including Boolean networks [5] [7], differential equations and Bayesian networks [8] [11], and vector autoregressive model [12].
We demonstrated that well-established tools of linear control theory can be used to model gene networks and explain experimental observations in a highly intuitive way.
State-of-the-art reverse-engineering methods model gene networks as static processes, i.e. regulatory interactions among genes in the network (such as direct physical interactions or indirect functional interactions) do not change across different conditions or tissue types.
Although we should exercise caution in comparing model gene networks with the segmentation clock (which additionally involves cell-signaling and dynamic complex formation), it seems likely that the presence of a greater number of periodically-expressed genes increases the opportunities for interaction turnover.
Each method is taken to run 10 times independently for modeling gene network, the threshold θ is set as 60%.
Application of the state-space equations to modeling gene networks allows us to use a large body of methodologies and tools in dynamic systems theory for studying dynamics of gene networks.
Several methods have been proposed for modelling gene networks including Boolean networks [ 7, 8], Bayesian networks [ 9], differential equations [ 10] and state space model (SSM) [ 11, 12].
Differential equation-based network models gene networks as a set of nonlinear differential equations that indicate the gene rate change without the assumption of discrete time steps [ 3].
For our discussion, these models are named the movement model, the gene network model and the full model.
