Pathway-Based Analysis Using SVM-RFE for Gene Selection and Classification

The pathway-based analysis is one method for selecting and classifying genes by incorporating pathway information. Integration of pathway knowledge into microarray data significantly advances researchers in the analysis of complex diseases. Microarray data involves thousands of genes to be selected, and therefore, a suitable method to eliminate noisy and uninformative genes is needed. Selecting significant genes for a specific disease is crucial to identifying genes highly related to disease production. Previous research shows that using both pathway information and gene expression data is more significant in disease identification. Therefore, pathway-based analysis using Support Vector Machine Recursive Feature Elimination (SVM-RFE) is introduced in this research to identify significant genes associated with analyzing the targeted phenotype. The datasets involved in this research are lung cancer and gender dataset. The results from the proposed method performed better than previous work as the significant genes are selected from the highest rank of genes in the highest rank of the pathway. The performance of the proposed method was evaluated using 10-fold cross-validation in terms of accuracy. Finally, a biological validation was conducted on selected genes in the top 5 pathways based on biological literature.