Optimizing Topology and Parameters of Gene Regulatory Network Models from Time-Series Experiments

Abstract

Abstract. In this paper we address the problem of finding gene regulatory networks from experimental DNA microarray data. Different approaches to infer the dependencies of gene regulatory networks by identifying parameters of mathematical models like complex S-systems or simple Random Boolean Networks can be found in literature. Due to the complexity of the inference problem some researchers suggested Evolutionary Algorithms for this purpose. We introduce enhancements to the Evolutionary Algorithm optimization process to infer the parameters of the non-linear system given by the observed data more reliably and precisely. Due to the limited number of available data the inferring problem is under-determined and ambiguous. Further on, the problem often is multi-modal and therefore appropriate optimization strategies become necessary. We propose a new method, which evolves the topology as well as the parameters of the mathematical model to find the correct network.

Multi-objective Model Optimization for Inferring Gene Regulatory Networks

“…4), the euclidian distance between the true system and the model found by the EA is getting worse, due to the sparseness of the true system. This validates the results of previous work [17]. The next two figures show the Pareto-front and the parameter distance of the 5-dimensional example in the case that the median average number of interactions of each gene was taken into account as the second optimization objective.…”

“…The ES optimizes the parameters of the mathematical model used for representation of the regulatory network. This algorithm was introduced by the authors in [17]. The current implementation is working on both, G and H, thus having the same number of bits as the algorithm in the previous case (2N 2 ).…”

Stages of Gene Regulatory Network Inference: the Evolutionary Algorithm Role

“…Reduction in the number of parameters to be optimised has been also performed using a nested optimisation approach, (Keedwell & Narayanan, 2005;Morishita et al, 2003;Spieth, Streichert, Supper, Speer & Zell, 2005;Spieth et al, 2004). These methods divide the search into two stages: structure and parameter search.…”

“…This reduces the number of real-valued parameters to be inferred at the second stage. The parameter search is performed using an evolution strategy, (Spieth, Streichert, Supper, Speer & Zell, 2005;Spieth et al, 2004), a genetic algorithm, , or back-propagation, ( Keedwell & Narayanan (2005), with an artificial neural network as the model). Again, this is facilitated by the flexibility of fitness evaluation, which is characteristic of evolutionary algorithms.…”

Multi-stage Evolutionary Algorithms for Efficient Identification of Gene Regulatory Networks

“…Evolutionary computation has been used from its inception for automatic identification of a given system or process [8]. For the S-system models, some evolutionary search techniques have been proposed [9][10][11][12]. However, they require the timeconsuming numerical integrations to reproduce dynamic profiles for the fitness evaluations.…”