Bioinformatics Research on the Investigation of Target Genes for Breast Cancer

Abstract

Cancer remains a worldwide problem, and so requires solutions. Research on cancer has been progressed extensively, but no perfect solution exists, despite the numerous diagnosis and treatment methods. Biomarkers are being used as a part of solutions against cancer, and research to discover new markers is being proceeded using many methods. In such research, bioinformatics methods can be used to find specific targets, with networks and gene ontology analysis relating to characteristics of cancer. Cancer has different features, and needs different approaches pathologically

thus, the present research was performed on breast cancer alone. First, database was constructed specializing in cancer and biomarkers for efficient management of data and for a firm foundation of analysis. Data required for the database construction were brought from many public repository such as NCBI, NCI, and UniProt. These data were processed by Html, Java, JavaScript, MySQL and JSP as several programming languages to fit their purposes. Second, cancer network was formed by utilizing the collected data. It was produced by mapping the cancer gene ID in protein interaction data, and inspected for power-law distribution of degree to see if the produced network has a scale-free nature. The cancer-specific network produced showed such a distribution, so the existence of the hub was ascertained, and reflecting on the characteristics of the graph determined the rank of hub genes. Accordingly, by calculating the network parameters such as degree distribution, betweenness centrality, and stress centrality, the top 150 hub genes were obtained. The following analysis filtered for specific GO terms reflecting on the characteristics of breast cancer through GO enrichment analysis utilizing 2,902 genes forming cancer network and breast cancer–related genes. By adjusting these specific GO terms to 2,902 genes of the cancer network, a gene list was obtained that is involved in general cancer with has breast cancer-specific GO terms. In the last step of the analysis, target genes were obtained through Venn diagram analysis by PPI network with GO enrichment analysis and target candidates, and results for five clinical markers were referred to. As a result, 34 genes with similar traits as clinical markers in PPI network and GO were selected as targets, and they are anticipated to be utilized primarily in selection of the next breast cancer marker. In addition, if such methods could filter candidates in selecting markers, they would be experimentally and clinically useful.