Influence of promoter length on network convergence in GRN-based evolutionary algorithms

Abstract

Genetic Regulation Networks (GRNs) are a model of the mechanisms by which a cell regulates the expression of its different genes depending on its state and the surrounding environment. These mechanisms are thought to greatly improve the capacity of the evolutionary process through the regulation loop they create. Some Evolutionary Algorithms have been designed to offer improved performance by taking advantage of the GRN mechanisms. A recent hypothesis suggests a correlation between the length of promoters for a gene and the complexity of its activation behavior in a given genome. This hypothesis is used to identify the links in in-vivo GRNs in a recent paper and is also interesting for evolutionary algorithms. In this work, we first confirm the correlation between the length of a promoter (binding site) and the complexity of the interactions involved on a simplified model. We then show that an operator modifying the length of the promoter during evolution is useful to converge on complex specific network topologies. We used the Analog Genetic Encoding (AGE) model in order to test our hypothesis. © 2011 Springer-Verlag.