Novel extension of k∈-∈TSP algorithm for microarray classification

Abstract

This paper presents a new method, referred as Weight k∈-∈TSP, which generates simple and accurate decision rules that can be widely used for classifying gene expression data. The proposed method extends previous approaches: TSP and k∈-∈TSP algorithms by considering weight pairwise mRNA comparisons and percentage changes of gene expressions in different classes. Both rankings have been modified as well as decision rules, however the concept of "relative expression reversals" is retained. New solutions to match analyzed datasets more accurately were also included. Experimental validation was performed on several human microarray datasets and obtained results are promising. © 2008 Springer-Verlag Berlin Heidelberg.