Gene Selection for Characterization of Differential Expressions in Microarray Experiments

Abstract

DNA microarray technology provides useful tools for profiling global gene expression patterns in different cell/tissue samples. One major challenge is the large number of genes relative to the number of samples. The use of all genes can suppress or reduce the performance of a classification rule due to the noise of none discriminatory genes. Selection of an optimal subset from the original gene set becomes an important pre-step in sample classification. Selection of an optimal gene set involves two steps: calculating a discriminatory index (score) for ranking gene with differential expressions and determining a cutoff from the ranked scores. This presentation considers several statistical discriminative indices based on the Receiver Operating Characteristic curve and uses the significance testing approach to determine a cutoff. We discuss the testing hypotheses behind these different statistical measures and the use of information of both the familywise error rate and false discovery rate as criterions for gene selection. Example data sets are used to illustrate an application of gene selection to sample classification. In addition, two classification algorithms, k nearest neighbors(kNN) and support vector machine (SVM), are used to evaluate the selected gene sets.