Estratégias evolutivas aplicadas a redes de regulação gênicas artificiais

Abstract

Gene regulatory networks (GRNs) influence the behavioral response of individuals when subjected to different contexts, they also affect extremely important processes for life, such as cell differentiation, metabolism and evolution. Computational models of gene regulatory networks, associated with artificial intelligence, enable us to create adaptable and context-independent solutions. In this work, we simulate the evolution of GRNs, aiming to evaluate how environmental variation and network growth events impact on the model's learning capacity. For this, we created populations of individuals represented by artificial gene regulatory networks (AGRNs), with physical characteristics and behaviors based on bacteria. We then simulated these populations on the tasks: “Objective Orientation”, “Phototaxy” and “Phototaxy with Obstacles”, evaluating how the events of single gene duplication, whole genome duplication and context change affect population evolution. The results indicated that a gradual increase in the complexity of the tasks performed is beneficial for the evolution of the model. Furthermore, we have seen that larger gene regulatory networks are needed for more complex tasks, with single-gene duplication being a good evolutionary strategy for growing these networks, as opposed to full-genome duplication. Studying how GRNs evolved in a biological environment allows us not only to improve the computational models produced, but also to provide insights into aspects and events that influenced the development of life on earth.