Review of the Extraction Methods of DNA Microarray Features Based on Central Decision Class Separation vs Rough Set Classifier

Abstract

AbstractThe study of DNA microarray gene extraction methods is an important and current area of research. Many researchers study gene ontological character, which contain significant information about symptoms of illnesses in tissues, types of organisms, and the distinguishing of some organisms’ features. DNA microarray gene extraction methods allow us to choose the most significant genes for a given problem and some ways of their extraction. In this article, we aim to compare three methods of gene extraction. The first and second types are based on, respectively, the modified Fisher and F statistics methods. The last one is based on the novel experimental statistics called A. A common element of those three methods is the way in which we choose genes after the calculation of decision classes’ separation ratio. Additionally, all three algorithms are based on the idea of central class separation from other decision concepts. We use our best 8v1.4 granular weighted voting classier as the basic element of comparison of our gene selection methods. The results of the research show that A statistics are better than other methods in all cases. In this article the best one is the SAM10 method, which works well for a small number of genes - less than one hundred. For a higher number of separated genes the SAM5 method is better - its effectiveness has been proven in recent published works.