Elham Salehi

The emergence of new genes in ecosim and its effect on fitness

The emergence of complex adaptive traits and behaviors in artificial life systems requires long term evolution with continuous emergence governed by natural selection. We model organisms genomes in an individual-based evolutionary ecosystem simulation (EcoSim), with fuzzy cognitive maps (FCM) representing their behavioral traits. Our system allows for the emergence of new traits and disappearing of others, throughout a course of evolution. We show how EcoSim models evolution through the behavioral model of its individuals governed by natural selection. We validate our model by examining the effect, the emergence of new genes, has on individuals fitness. Machine learning tools showed great interest lately in modern biology, evolutionary genetics and bioinformatics domains. We use Random Forest classifier, which has been widely used lately due to its power of dealing with large number of attributes with high efficiency, to predict fitness value knowing only the values of new genes. Furthermore discovering meaningful rules behind the fitness prediction encouraged us to use a pre processing step of feature selection. The selected features were then used to deduce important rules using the JRip learner algorithm.

Efficient EDA for large opimization problems via constraining the search space of models

Introducing efficient Bayesian learning algorithms in Bayesian network based EDAs seems necessary in order to use them for large problems. In this paper we propose an algorithm, called CMSS-BOA, which uses a recently introduced heuristic called max-min parent children (MMPC) [3] in order to constraint the models search space. This algorithm does not consider a fix and small upper bound on the order of interaction between variables and is able solve problems with large number of variables efficiently. We compare the efficiency of CMSS-BOA with standard Bayesian network based EDA for solving several benchmark problems.

Estimation of Distribution Algorithms in Gene Expression Data Analysis

Estimation of Distribution Algorithm (EDA) is a relatively new optimization method in the field of evolutionary algorithm. EDAs use probabilistic models to learn properties of the problem to solve from promising solutions and use them to guide the search process. These models can also reveal some unknown regularity patterns in search space. These algorithms have been used for solving some challenging NP-hard bioinformatics problems and demonstrated competitive accuracy. In this chapter, we first provides an overview of different existing EDAs and then review some of their application in bioinformatics and finally we discuss a specific problem that have been solved with this method in more details.