Comparing Genomic Network Methodologies: A Combined Approach for Cancer Prognosis

摘要

One of the goals of cancer research is to understand the genetic causes of disease pathology and specify the exact ways that genetic components interact to enable a complex living system exhibit the disease phenotype. Consequently, research efforts must be addressed to elucidate various phenome components, such as trancriptome, metabolome and proteome, with the aim to derive a prognostic phenotype. In this work, we attempt to model causal effects among genes and proteins using their interactions in the form of biological networks. Two spatial network approaches are examined in breast cancer in association with established genomic signatures, in order to derive tight subnetworks linked to explicit biological processes. These approaches include the HotNet2 and Activity Vector algorithms, which create gene interaction subnetworks after processing and evaluating gene expression data. Finally, we evaluate the results for their biological significance and their statistical prediction in an independent dataset. The proposed network analysis provides a blueprint to explore diagnostic and/or therapeutic opportunities.