Wagner's canalization model
Introduction
The idea that wild-type genotypes are mutationally robust, i.e., are buffered against the effect of mutation goes back to Waddington (1942), an effect he called canalization. Waddington argued that for a well adapted trait almost all mutations with an effect are deleterious. For this reason, any modifier that reduces the effect of mutations and thus keeps the trait closer to optimum should be selected. Following the classical work of Schmalhausen (1949), Waddington (1957) and their contemporaries, research on robustness experienced a decline in the 1970s and 1980s. However, in the early 1990s as powerful molecular techniques to track and manipulate genotypes became routine, there was a renewed interest in the issue of genetic robustness.
Scharloo (1991) describes the results of a number of early experiments with Drosophila. More recently Rutherford and Lindquist (1998) and Queitsch et al. (2002) have examined the role of the heat shock protein Hsp90. All these experiments show increased phenotypic variance in populations that carry a major mutation or are exposed to environmental stress and this is interpreted as evidence for the existence of an evolutionary buffer capable of hiding and then releasing genetic variation. See Stearns (2002) and Niven (2004) for more discussion of the experiments.
The evolution of mechanisms underlying the buffering of the phenotype against genetic and environmental influences has received much theoretical attention, yet many issues remain unresolved. For discussions see Meiklejohn and Hartl (2002), Gibson and Wagner (2000), or De Visser et al. (2003). There have been a number of different approaches to modeling, see e.g., Hermisson et al. (2003), Hermisson and Wagner (2004). This paper follows the approach of Wagner (1996) and Siegal and Bergman (2002) who investigated an interacting network of N genes described by an matrix where on and off . Masel (2004) later investigated a variant in which on and off.
In Section 2 we will describe the original version of Wagner's model and Masel's modification of the model. Section 3 introduces three population dynamics: one due to Wagner (1996), one to Siegal and Bergman (2002), and an intermediate model which is a hybrid of the two. Section 4 is devoted to a detailed study of the case and a use of Markov chain theory to derive a result for the asymptotic behavior in an infinite population for a general N. In Section 5, we investigate systems with and 10 genes by simulation. In Section 6, we discuss the conclusions we have reached based on our analytical results and simulations.
Section snippets
The network model
Following Wagner (1996) we consider a finite population of M randomly mating individuals, each of which has an interacting network of N genes. These interactions are represented by an matrix W, whose elements indicate the effect on gene i of the product of gene j, which may involve activation or repression . Changing expression levels are modeled by the set of difference equationswhere is the sign function; if , if , and
Population dynamics
For simplicity, we will describe the simulations for the case. We consider three different evolutionary simulations. For all three, we assume the Wagner mapping for the network dynamics and for comparison in one of the scenarios we apply the 0, 1 map.
In the first simulation, which is similar to the approach of Wagner (1996), we have a fixed optimum phenotype, which without loss of generality we can suppose is . To generate the founding population, we generate 10,000 random
Markov chain model
To gain further insight into the structure of the Wagner map, we will look at the simplest nontrivial case. In the we can explicitly draw out what the trajectories on the state space will look like. Fig. 2 has all the possible types of systems and how many of each type occur (by permuting the positions of orbits and fixed points while still keeping the system antisymmetric). The antisymmetry is geometrically an invariance through the center of the square whose corners represent the
Path length and probability of viability for
In higher dimensions it is impossible to visualize the distribution of even one row, so we instead investigate statistics associated with the mapping. We compute the mean path length at each time point for each run and then take the average over the number of runs. Path length is the number of steps that the individual takes to reach an equilibrium (fixed point), and a run simply means that a population of networks has been evolved for a fixed number of generations which is 200 for the first
Discussion
We have examined a model of the evolution of a population of genetic networks that extends earlier work of Wagner (1996) and Siegal and Bergman (2002), who investigated particular cases of the model by simulation. The aim of this investigation is to obtain a more thorough understanding of the properties of the model, in order to better understand the conclusions that can be drawn from the observed behavior. As in previous work, we find that the networks evolve to be more robust. We find these
References (21)
- et al.
A single mode of canalization
Trends Ecol. Evol.
(2002) - et al.
Plasticity, evolvability, and modularity in RNA
J. Exp. Zool.
(2000) - et al.
Sexual reproduction selects for robustness and negative epistasis in artificial gene networks
Nature
(2006) - et al.
Evolutionary capacitance as a general feature of complex gene networks
Nature
(2002) Perspective: evolution and detection of genetic robustness
Evolution
(2003)- et al.
Canalization in evolutionary genetics: a stabilizing theory?
BioEssays
(2000) - et al.
The population genetic theory of hidden variation and genetic robustness
Genetics
(2004) - et al.
Epistasis in polygenic traits and the evolution of genetic architecture under stabilizing selection
Am. Nat.
(2003) Genetic assimilation can occur in the absence of selection for the assimilating phenotype, suggesting a role for the canalization heuristic
J. Evol. Biol.
(2004)Channeling evolution: canalization and the nervous system
PLoS Biol.
(2004)
Cited by (24)
Using evolutionary computations to understand the design and evolution of gene and cell regulatory networks
2013, MethodsCitation Excerpt :Wagner [25] and Siegal and Bergman [26], simulating N-gene network evolution, have argued that networks evolve to be robust. Huerta and Sanchez-Durrett [32], by varying system size and convergence criteria, with and without recombination, reported that this evolution to robustness is itself a conserved trait, i.e. not dependent on the details of the model. Leclerc [36] introduced measures for the cost of GRN complexity to the Wagner approach.
Phenotype-genotype relation in Wagner's canalization model
2012, Journal of Theoretical BiologyCitation Excerpt :They report that the evolution of robustness goes along with an increase in phenotypic diversity revealed in response to noise. Finally, in terms of theoretical analysis, Huerta-Sanchez and Durrett (2007) have proposed a mathematical framework for a simplified version of the WM. In the case of discrete mutations and no development dynamics, they have shown that robustness also lies in the exploration of the phenotypic space. Because of the threshold function, the viable genotypes can be located in the genotype space, and mutations may drive them outside of the viability zone.
Robustness and evolvability
2010, Trends in GeneticsCitation Excerpt :In the case of protein sequences, this bimodality might often correspond to the destruction of a protein fold versus tinkering with function within the same overall fold. Simple gene regulatory networks can also be analyzed as neutral networks [62–64]. Gene regulatory networks show immense phenotypic richness in the neighborhood of a typical neutral network [62].
Chaotic gene regulatory networks can be robust against mutations and noise
2008, Journal of Theoretical BiologyEffective models for gene networks and their applications
2012, Biophysical Reviews and Letters
- 1
Supported by a Cornell-Sloan fellowship.
- 2
Partially supported by NSF grants from the probability program (0202935) and from a joint DMS/NIGMS initiative to support research in mathematical biology (0201037).