Investigating the minimum required number of genes for optimum classification of myopathy microarray data

The investigation of potential microarray markers, which in turn will speed up the molecular analysis and provide reliable results on the benefit of patient care is of significant importance. Feature selection techniques, which aim at minimizing the dimensionality of the microarray data by keeping the most significant genes according to their expression values is a necessary component towards this goal. In the current article, we present an investigation regarding the minimum required subsets of genes, which best classify myopathy data. For this purpose, we developed a tool that facilitates the users to easily access/use multiple feature selection methods and subsequently perform classification of data. For the current study, five feature selection methods on datasets from two different myopathies have been utilized. Our findings reveal subsets of very small number of genes, which can successfully classify gene expression datasets from different patients with skeletal myopathies. In addition, we observe that similar classification results may be obtained from completely different subsets of genes. The developed tool can expedite the identification of small gene subsets with high classification accuracy that could ultimately be used in the genetics clinics for diagnostic, prognostic and pharmacogenomic purposes.