TU Darmstadt / ULB / TUbiblio

Evolution of Boolean networks under selection for a robust response to external inputs yields an extensive neutral space

Szejka, Agnes and Drossel, Barbara (2010):
Evolution of Boolean networks under selection for a robust response to external inputs yields an extensive neutral space.
In: Physical Review E, pp. 021908, 1-9, 81, (2), ISSN 1539-3755, [Online-Edition: http://dx.doi.org/10.1103/PhysRevE.81.021908],
[Article]

Abstract

We study the evolution of Boolean networks as model systems for gene regulation. Inspired by biological networks, we select simultaneously for robust attractors and for the ability to respond to external inputs by changing the attractor. Mutations change the connections between the nodes and the update functions. In order to investigate the influence of the type of update functions, we perform our simulations with canalizing as well as with threshold functions. We compare the properties of the fitness landscapes that result for different versions of the selection criterion and the update functions. We find that for all studied cases the fitness landscape has a plateau with maximum fitness resulting in the fact that structurally very different networks are able to fulfill the same task and are connected by neutral paths in network (“genotype”) space. We find furthermore a connection between the attractor length and the mutational robustness, and an extremely long memory of the initial evolutionary stage.

Item Type: Article
Erschienen: 2010
Creators: Szejka, Agnes and Drossel, Barbara
Title: Evolution of Boolean networks under selection for a robust response to external inputs yields an extensive neutral space
Language: English
Abstract:

We study the evolution of Boolean networks as model systems for gene regulation. Inspired by biological networks, we select simultaneously for robust attractors and for the ability to respond to external inputs by changing the attractor. Mutations change the connections between the nodes and the update functions. In order to investigate the influence of the type of update functions, we perform our simulations with canalizing as well as with threshold functions. We compare the properties of the fitness landscapes that result for different versions of the selection criterion and the update functions. We find that for all studied cases the fitness landscape has a plateau with maximum fitness resulting in the fact that structurally very different networks are able to fulfill the same task and are connected by neutral paths in network (“genotype”) space. We find furthermore a connection between the attractor length and the mutational robustness, and an extremely long memory of the initial evolutionary stage.

Journal or Publication Title: Physical Review E
Volume: 81
Number: 2
Divisions: 05 Department of Physics
05 Department of Physics > Institute for condensed matter physics
05 Department of Physics > Institute for condensed matter physics > Statistische Physik und komplexe Systeme
Date Deposited: 26 Oct 2010 08:28
Official URL: http://dx.doi.org/10.1103/PhysRevE.81.021908
Identification Number: doi:10.1103/PhysRevE.81.021908
Export:

Optionen (nur für Redakteure)

View Item View Item