Link-based identification of survival time-related biological pathways

This paper proposes a novel survival time pathway hunting method based on gene links. In the method, we incorporate gene link information for testing how significantly a pathway is associated with cancer patient´s survival. Specifically, we establish a link-based Cox proportional hazard model (Link-Cox), in which two linked genes are considered together as a link variable and the association with survival time is assessed using Cox proportional hazard model. Based on the Link-Cox model, A new statistic for measuring the association of a pathway with survival time, named pathway survival score (PSS), is formulated by summarizing survival significance over all the gene links in the pathway. We also devise a permutation test to estimate the significance of an observed PSS. To evaluate the proposed method, we applied it to two publicly available lung cancer gene expression data sets. Experimental results on two real datasets show the effectiveness and efficiency of the proposed method.