Metaheuristic optimization in Systems Biology

Community detection has been designed as an interesting and valuable approach in Complex Network Analysis. Constitutes an important field of research were complex network represent a powerful interpretation tool widely involved in neuroscience and biology, social science, economy and many others. Uncovering the community structure is a crucial step towards a better understanding of complex systems, revealing the internal organization of nodes. In recent literature several methods has been proposed in order to solve problem of Community Detection, and many efforts were made to develop a well-functioning clustering algorithm. However, existing algorithms in the literature up-to-date present several crucial issues, and the question of how good an algorithm is, with respect to others, is still open. In this thesis, we carry out an exploration of the existing techniques for the Community Detection problem, following which we propose an approach based on the measure of Modularity suggested by Newman. This turns out to be a natural estimate of the goodness of the network community decomposition and is one of the widely used methods, with the contribution of metaheuristic methods, to address optimization problems in community determination. We investigate the link between biological modules and network communities in a test-case networks that are commonly used as a reference point. Through the study of biological networks, such as Protein Protein Interactions, it is possible to identify the intrinsic structure of molecular interactions, in order to identify the so-called "disease modules", consisting of a set of genes that often have a similar biological role. Identifying these modules would open up new perspectives for therapeutic applications such as targeted drug development. While the topological analysis can give insight on internal organization of nodes, investigation influences of single nodes in a biological network has a potential of practical applicability such as the identification of drug targets. We identify our problem as a maximum flow problem by evaluating the amount of flows that can pass between candidate proteins and disease genes. In this thesis, we explore computational strategies for the maximum flow problem with the aim to identify important nodes in the network.