Gene Selection for Colon Cancer Classification using Bayesian Model Averaging of Linear and Quadratic Discriminants

Abstract

Recent findings reveal that various cancer types can be diagnosed using non-clinical approach which involves monitoring of the biological samples using their genes expression profiles. Two of the widely used methods are Linear and Quadratic discriminant analyses. In this paper, the behaviours of Linear and Quadratic Discriminants Analyses (LDA and QDA) were observed within the framework of Bayesian model averaging. We applied Bayesian model averaging to tackle model uncertain problem inherent in discriminant analysis. We calibrated the developed classifier on published real life microarray colon cancer data that contained 2000 gene expression profiles measured on 62 biological samples that comprised 40 tumorous tissue samples and 22 normal tissue samples. The data were also pre-processed using logarithmic transformation to base 10 and zero mean unit variance normalization as often done with the dataset. In addition, comparison with Random Forest (RF), Gradient Boosting Machine (GBM) and Bayesian Additive Regression Trees (BART) was also achieved. Various performance results established the supremacy of the proposed method.