Briefings in Bioinformatics Advance Access published online on September 29, 2009
Briefings in Bioinformatics, doi:10.1093/bib/bbp035
Advances in metaheuristics for gene selection and classification of microarray data
Corresponding author: Jin-Kao Hao, University of Angers, 2 Boulevard Lavoisier, 49045 Angers Cedex 01, France. Tel: +33-241-735076; Fax: +33-241-735073, E-mail: hao{at}info.univ-angers.fr
Gene selection aims at identifying a (small) subset of informative genes from the initial data in order to obtain high predictive accuracy for classification. Gene selection can be considered as a combinatorial search problem and thus be conveniently handled with optimization methods. In this article, we summarize some recent developments of using metaheuristic-based methods within an embedded approach for gene selection. In particular, we put forward the importance and usefulness of integrating problem-specific knowledge into the search operators of such a method. To illustrate the point, we explain how ranking coefficients of a linear classifier such as support vector machine (SVM) can be profitably used to reinforce the search efficiency of Local Search and Evolutionary Search metaheuristic algorithms for gene selection and classification.
Keywords: microarray data analysis, gene selection, classification, local search, genetic algorithm, memetic algorithm
Submitted: May 5, 2009. Received (in revised form): July 12, 2009.