Abstract
Many biological processes, at any level of organization from cellular to ecosystem, can be modeled as a complex network. In ecosystems, the objects in the network are the organisms involved in the model and the relationships are how the organisms interact with each other. At the cellular level, objects range from genes to metabolites and the relationship represent the interactions between them. In recent years, we have witnessed an explosion of available biological data that started with the Human Genome Project, and then it matured with the birth of a new field of study called systems biology. In this field, available data is integrated and viewed from the systems perspective. A vast amount of data has become publicly available, and a significant amount of it can be modeled using networks. A subfield of systems biology, network biology takes special interest in the biological network models. Network biology helps the biomedical community to unravel the mysteries of life, and also of diseases. Identification of diseases is a long time research subject, in this thesis we will present a method for disease gene identification using biological networks approach. The approach is based on protein interaction networks and microarray expression data. It integrates techniques like random walk with restarts with filtering purposes, shortest paths analysis for the core of the prioritization, and topological features of the network to help identify key genes. The contribution of this thesis is an integrated method for disease gene prioritization, that was tested using the prostate cancer as domain, obtaining the best performance for the top 50 rank compared to other state of the art methods.