Biomarker identification by feature wrappers

Gene expression studies bridge the gap between DNA information and trait in formation by dissecting biochemical pathways into intermediate components b etween genotype and phenotype. These Studies open new avenues for identifyi ng complex disease genes and biomarkers for disease diagnosis and for asses sing drug efficacy and toxicity. However, the majority of analytical method s applied to gene expression data are not efficient for biomarker identific ation and disease diagnosis. In this paper, we propose a general framework to incorporate feature (gene) selection into pattern recognition in the pro cess to identify biomarkers. Using this framework, we develop three feature wrappers that search through the space Of feature subsets using the classi fication error as measure of goodness for a particular feature subset being "wrapped around": linear discriminant analysis, logistic regression, and s upport vector machines. To effectively carry Out this computationally inten sive search process, we employ sequential forward search and Sequential for ward floating search algorithms. To evaluate the performance of feature sel ection for biomarker identification we have applied the proposed methods to three data sets. The preliminary results demonstrate that very high classi fication accuracy can be attained by identified composite classifiers with several biomarkers.