How Can Problem Structure Analysis from Evolutionary Computation be Applied to Biological Problems?

Presenter: Robert Heckendorn, Department of Computer Science, University of Idaho

Abstract:

In Biology, understanding the interaction of genes with other genes and their environment is critical to understanding how organisms function and evolve. In Computer Science, understanding the "structure" of a problem representation is critical to understanding how to effectively solve an optimization problem using evolutionary computation. While these two problems are very similar, there are some severe practical differences in these two fields. In this talk I will introduce epistasis concepts and briefly explore how the fitness landscapes the populations evolve on influence their trajectories in gene space. I will also discuss the practical effort to adapt the analysis techniques found in Evolutionary Computation to the needs of biologists. I look at some work in a new grant in which we are developing specific algorithms in collaboration with researchers at BEACON Center for Evolution in Action at Michigan State and give some examples from real biological problems.