A multifuzzy filtering approach to reliable gene expression profile analysis

Computational intelligent methods have become popular tools for gene expression profile analysis and for identifying a set of significant genes. However, their conclusions rely on an outcome of a single method. This work highlights the fact that each method can result in a different gene profile and set of the most significant genes. Therefore, the result cannot be accurate, and most importantly, not reliable. In order to draw a more reliable decision, a multifuzzy filtering approach to gene expression profile analysis is developed. The method is applied to the analysis of the yeast genes. The results suggest that the genes identified are more reliable and provide more valuable information than those stratified by a single method.