Identifying module biomarkers of hepatocellular carcinoma from gene expression data

Identifying effective cancer biomarkers is crucial in precision medicine. Based on the high-throughput available omics data such as microarray, this paper aims to identify potential biomarker genes for hepatocellular carcinoma by bioinformatics and machine learning. We describe the gene coexpressions with network model and detect out the genes that are closely related to liver cancer infected by hepatitis virus. We cluster these genes by the network topology and then evaluate their classification performance of distinguishing controls from disease samples by support vector machine classification. The functional enrichments of the gene group are also implemented and analyzed. These genes with good classification power and dysfunctional implications are identified as candidate biomarkers for hepatocellular carcinoma.