Gene Regulatory Network Evolution Through Augmenting Topologies

Abstract

Artificial gene regulatory networks (GRNs) are biologically inspired dynamical systems used to control various kinds of agents, from the cells in developmental models to embodied robot swarms. Most recent work uses a genetic algorithm (GA) or an evolution strategy in order to optimize the network for a specific task. However, the empirical performances of these algorithms are unsatisfactory. This paper presents an algorithm that primarily exploits a network distance metric, which allows genetic similarity to be used for speciation and variation of GRNs. This algorithm, inspired by the successful neuroevolution of augmenting topologies algorithm's use in evolving neural networks and compositional pattern-producing networks, is based on a specific initialization method, a crossover operator based on gene alignment, and speciation based upon GRN structures. We demonstrate the effectiveness of this new algorithm by comparing our approach both to a standard GA and to evolutionary programming on four different experiments from three distinct problem domains, where the proposed algorithm excels on all experiments.

SNT: A Unifying Toolbox for Quantification of Neuronal Anatomy

“…The quantitative measurements made by SNT can be used to statistically test the difference between morphological populations of cells. We demonstrate this approach by randomly generating five unique gene regulatory networks (GRNs) 60 that satisfy minimal viability criteria. A simple test for viability was performed as follows: the GRN is simulated for 100 time-steps with constant input values, with the following criteria assessed at the first and final time-step of the viability test: (1) all behavior regulating protein concentrations should change, (2) the GRN must bifurcate enough to have reasonable growth, (3) the GRN must not over-bifurcate, (4) the GRN must trigger branching, and (5) the GRN must not overbranch.…”

Evolved Developmental Strategies of Artificial Multicellular Organisms

“…newly produced proteins and the destroyed one. More details on the aGRN dynamics can be found in (Cussat-Blanc et al, 2015). Table 1 describes the configuration of our aGRN input and output proteins when applied to this artificial embryogenesis problem.…”

“…The first employed approach was to use a standard objective based genetic algorithm (GA). We implemented most features of the Gene Regulatory Network Evolution through Augmenting Topology algorithm (GRNEAT) (Cussat-Blanc et al, 2015).…”

Genetically-regulated Neuromodulation Facilitates Multi-Task Reinforcement Learning

“…In this paper, we use an evolutionary algorithm designed specifically for the evolution of GRNs called GRNEAT [3]. GRNEAT incorporates three key features: initialize with small networks, speciation of GRNs based upon a measure of similarity between networks, and an aligned crossover that preserves subnetwork architecture when recombining GRNs.…”

“…This modified algorithm has been shown to converge faster and to better solutions than standard genetic algorithms. More details on GRNEAT can be found in [3].…”