Specific tuning parameter for directed random walk algorithm cancer classification

Accuracy of cancerous gene classification is a central challenge in clinical cancer research. Microarray-based gene biomarkers have proved the performance and its ability over traditional clinical parameters. However, gene biomarkers of an individual are less robustness due to litter reproducibility between different cohorts of patients. Several methods incorporating pathway information such as directed random walk have been proposed to infer the pathway activity. This paper discusses the implementation of group specific tuning parameter in directed random walk algorithm. In this experiment, gene expression data and pathway data are used as input data. Throughout this experiment, more significant pathway activities can be identified which increases the accuracy of cancer classification. The lung cancer gene is used as the experimental dataset, with which, the sDRW is used in determining significant pathways. More risk-active pathways are identified throughout this experiment.