New approaches to designing genes by evolution in the computer
Holloway, David (David Holloway (David_Holloway)) (author)
Spirov, Alexander, V. (author)
© 2012 InTech
The field of Evolutionary Computation (EC) has been inspired by ideas from the classical theory of biological evolution, with, in particular, the components of a population from which reproductive parents are chosen, a reproductive protocol, a method for altering the genetic information of offspring, and a means for testing the fitness of offspring in order to include them in the population. In turn, impressive progress in EC - understanding the reasons for efficiencies in evolutionary searches - has begun to influence scientific work in the field of molecular evolution and in the modeling of biological evolution (Stemmer, 1994a,b; van Nimwegen et al. 1997; 1999; Crutchfield & van Nimwegen, 2001). In this chapter, we will discuss how developments in EC, particularly in the area of crossover operators for Genetic Algorithms (GA), provide new understanding of evolutionary search efficiencies, and the impacts this can have for biological molecular evolution, including directed evolution in the test tube. GA approaches have five particular elements: encoding (the ‘chromosome’); a population; a method for selecting parents and making a child chromosome from the parents' chromosomes; a method for altering the child’s chromosomes (mutation and crossover/recombination); criteria for fitness; and rules, based on fitness, by which offspring are included into the population (and parents retained). We will discuss our work and others’ on each of these aspects, but our focus is on the substantial efficiencies that can be found in the alteration of the child chromosome step. For this, we take inspiration from real biological reproduction mechanisms.