Algoritmos genéticos para predição ab initio de estrutura de proteínas

• Ano de defesa: 2008
• Autor(a) principal: Custódio, Fábio Lima
• Orientador(a): Dardenne, Laurent Emmanuel
• Banca de defesa: Galeão, Augusto Cesar Noronha Rodrigues , Borges, Carlos Cristiano H. , Raupp, Fernanda Maria Pereira , Bisch, Paulo Mascarello , Garratt, Richard Charles
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Laboratório Nacional de Computação Científica
• Programa de Pós-Graduação: Programa de Pós-Graduação em Modelagem Computacional
• Departamento: Serviço de Análise e Apoio a Formação de Recursos Humanos
• País: BR
• Palavras-chave em Português: Biologia molecular > Processamento eletrônico de dados; Algorítmos genéticos; Predição de estrutura de proteínas; Modelo HP
• Palavras-chave em Inglês: Molecular biology > Electronic data processing; Genetic algorithms
• Área do conhecimento CNPq: CNPQ::CIENCIAS BIOLOGICAS::BIOQUIMICA::BIOLOGIA MOLECULAR
• Link de acesso: https://tede.lncc.br/handle/tede/89

The goal of ab initio protein structure prediction (PSP) methods is to predict, based on first principles, the three-dimensional structure that a given amino-acid sequence will assume. These methods have important biotechnological applications from the creation of new proteins, drug design (the receptor's structure), the refinement of theoretical models and the elucidation of structures from sparse experimental data. The prediction encompasses an optimization problem with thousands of degrees of freedom and is associated with extremely complex energy hypersurfaces. This results in a problem that is difficult to treat computationally. In this work a simplified three-dimensional protein model (hydrophobic-polar model, HP) was used in order to reduce the computational costs of the PSP problem allowing for the fast development of a robust and efficient genetic algorithms based methodology. This methodology was then adapted to an all-atoms protein model. During the development different strategies for the PSP problem were analyzed. A new crowding based approach is described for the maintenance of diversity within the population which resulted in achieving multiple solutions. The HP model's methodology was tested against all 35 major sequences from the literature and the comparative results showed that the proposed genetic algorithm is superior to other evolutionary algorithms and comparable to specialized methods. The algorithm applied to the all-atoms model was initially tested with poli-alanines and later with five other proteins. Structures with RMSDs ranging from 2.0 to 6.7 Å were found and the proposed algorithm was superior to others similar methods, in terms of computational costs. The results obtained showed that optimization strategies with multiple solutions characteristics present two advantages. The first one is the more efficient investigation of a complex hypersurface, with better chances of finding optimal solutions; the second one is increasing the probability of finding structures close to those experimentally determined, even when they are not near the global optimum of the energy hypersurface.