DISCOVER GENE ASSOCIATION IN COLON CANCER DISEASE USING SUPPORT VECTOR MACHINE AND MULTI-OBJECTIVE OPTIMIZATION TECHNIQUE

Abstrak

gene expression to detect association of gene in cancer disease is still promising. Each gene have dominant and less dominant impact to construct a disease. To analyze the cause and prevent further spread of cancer tissue, gene disease association based on gene expression is needed to discover correlation between gene that cause cancer and does not have correlation to cancer. Commonly, to discover gene association in colon cancer disease is lack of patient and number of gene that scrambled between up-regulated and down-regulated to this disease. In this study, Support Vector Machine conducted to separate between mutated colorectal cancer (MCC) and normal patient. Recursive feature elimination (RFE) is conducted to rank and select most informative gene. Furthermore, multi-objective optimization technique named non dominated sorting genetic algorithm (NSGA-II) is applied to find minimal solution of gene selection. To validate classification of gene expression data, Support vector machine obtain 99.6% of accuracy, while precision reach 100% and 98.6% of recall.