Predição computacional de interações de proteína-proteína em proteomas preditos de Leishmania

• Ano de defesa: 2012
• Autor(a) principal: Antônio Mauro Rezende
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Bioinformática; Proteoma; Leishmania; Proteínas
• Link de acesso: http://hdl.handle.net/1843/BUOS-998GUB

The Trypanosomatid parasites Leishmania braziliensis, Leishmania infantum and Leishmania major are important human pathogens. Despite years of study and the availability of their genomes, no effective vaccine was developed until the present moment, and the available treatments are in general highly toxic. Therefore, it is clear that only integrated studies with an interdisciplinary approach will be succeeded intrying to search new targets for drug and vaccine development. One essential part of this rational is related to protein-protein interactionnetwork (PPI) study which can provide a better understanding of complex protein interactions in biological systems. Thus, in the present doctorate thesis, we modeled PPI for the three above cited species of Leishmania by the computational method that uses sequence comparison approach (Interolog Mapping), and developed a system of combined score to evaluate the robustness of the predictions. The performance evaluation of the PPI prediction approach was performed using a set of protein interaction data of Escherichia coli as gold standard, and the value of AUC found was 0.94. As result, 39,420, 45,325, and 43,531 interactions were predicted for L. braziliensis, L. infantum and L. major, respectively. For each PPI predicted, the top 20 ranked proteins in according to topological index MCC (Maximal Centrality Clique) were selected. Furthermore, information related to the conservation of protein sequence among orthologs, level of identity compared to potential host proteins, and immunological potential were integrated. Here, it is worth highlighting that the algorithms used to epitope prediction had their performances previously evaluated. This was performed utilizing data from IEDB as gold standard. Hence, the programs with the best performance were employed. Therefore, this integration provides a better understanding and usability of the PPIs predicted which can be valuable for selection of new biological targets for drug and vaccine development.Other point that deserved attention in this study is linked to network modularity, focusing on conserved modules, key feature when one is interested in destabilizing the PPI for drug and vaccine purpose. These analyses revealed a pattern associated with protein turnover. In addition, nearly 50% of the proteins describes as hypothetical present in the PPIs received some level of functional annotation, which represent an important contribution since approximately 60% of predicted proteome of species from Leishmania genus does not have any functional prediction.