A comparision between methods for generating differentially expressed genes from microarray data for prediction of disease

Abstract

Feature selection from microarray data has become an ever evolving area of research. Numerous techniques have widely been applied for extraction of genes which are expressed differentially in microarray data. Some of these comprise of studies related to fold-change approach, classical t-statistics and modified t-statistics. It has been found that the gene lists returned by these methods are dissimilar. In this work we compare the outputs of two different feature selection methods using three classifiers based on different algorithms namely the Random Forest Ensemble based method, the Support vector machine (SVM) and the KNN methods, using the prediction accuracy of the test datasets.