Um modelo para a evolução de redes de proteínas contemplando a transferência horizontal de genes

Abstract

Around us we find several examples of complex systems, which are composed of a large number of elements that interact with each other in very complex ways. These systems range from the World Wide Web itself to energy distribution networks and protein interaction networks (PIN), which are the focus of this work. It is possible to represent the PINs, as well as other complex systems, through a set of nodes and links, forming a graph from which we can extract the properties of interest to study these networks. Among the wide variety of properties that we can use to study PINs, we focused in trying to understand the evolutionary processes that are responsible for originating the networks as we know them today. Thus, some networks were selected and from them we determined their topology and eigenvalue spectrum, in order to have comparison tools. After choosing the networks, we searched the literature for models that had already been developed in order to reproduce such networks, where we found the Barabási and Albert (BA) model and the Duplication-Divergence (DD) model. Despite being well-regarded models, they are not capable of reproducing all the relevant properties of the PINs, so that new evidence was sought for the development of a model that was more accurate and produced better results. Some studies suggests the existence of another mechanism that alongside gene duplication is responsible for the evolution of PINs: the horizontal gene transfer (HT). However, we were not able to find evidences about how this mechanism really works and how we can implement it. Thus, starting from the hypothesis that the new proteins added to an organism through HT will have a greater preference to connect and interact with the most connected proteins already existing in the network, we decided to use a modified version of the BA model to represent the HT mechanism. We then developed a new model for the evolution of PINs which is based in these two mechanisms: duplication and horizontal transfer. As we can see throughout this work, when we analyze the topology, eigenvalue spectrum and other properties of the networks generated with the new model, the results are closer to the real networks when comparing them to the results generated with the DD model, which supports and strengthen the idea that gene duplication is not the only mechanism behind the evolutionary proccess of protein interaction networks.