Investigating survival prognosis of glioblastoma using evolutional properties of gene networks

In recent years, there has been widespread interest and a large number of publications on the application of graph theory techniques into constructing and analyzing biologically-informed gene networks from cancer cell line data sets. Current research efforts have predominantly looked at an overall static, topological, representation of the network, and have not investigated the application of graph theoretical techniques to evolutionary investigations of cancer. A number of these studies have used graph theory metrics, such as degree, betweenness, and closeness centrality, to identify important hub genes in these networks. However, these have not fully investigated the importance of genes across the different stages of the disease. Previous human glioblastoma publications have identified four subtypes of glioblastoma in adults, based on signature genes. In one such publication, Verhaak et al. found that the subtypes correspond to a narrow median survival range, from 11.3 months for the most aggressive subtype, to 13.1 months for the least aggressive one. In this work, we present an evolutionary graph theory study of glioblastoma based on survival data categorization, confirming genes associated with different survival times identified using established graph theory metrics. The work is extending the application of graph theory approaches to evolutionary studies of cancer cell line data.