Abstract:
|
Ovarian cancer is the fifth leading cause of cancer death. In this study, our goal is to integrate multiple omics data sets using network analysis tools to derive gene expression signatures useful to predict survival outcomes. We construct a regulatory network of CNAs and mRNA expressions based on genomic profiles of 293 ovarian cancer samples from TCGA using a newly developed statistical tool, spaceMap, a conditional graphical model, which learns the conditional dependency relationships between two types of nodes through a penalized multivariate regression framework. Based on the network, we identify a collection of hub genes that are influenced directly by CNA and are interacting with a large number of other genes. We then select a subset of these genes whose expressions also appeared to have a big impact on global protein activities based on both mRNA and proteomics profiles of 105 ovarian tumors from CPTAC study. We expect such a strategy would lead to better prediction models and novel drug targets to improve the patient disease outcome.
|