Combining Network Topology With Transcriptomic Data For Identifying Radiosensitive Gene Signatures

Radiotherapy sensitivity assay can gain insight into radiation biology and clinically improve radiotherapy of cancer patients. This paper proposesd to combine gene network information and gene expression profiles for identifying radiosensitive signatures. Specifically, the association of each gene with survival fraction (SF) induced by radiation was first estimated using Spearman correlation measure and then adjusted based on network topology. We used directed random walk (DRW) algorithm to assess the topological importance of genes. Two types of molecular networks, protein-to-protein interaction network and gene network, were considered in the DRW algorithm. The identified signatures were verified via pathways enrichment analysis, shown rich in disease pathways. The findings suggested useful guidelines for the clinical practice of radiotherapy of cancer patients.