Resolução de estruturas de proteínas utilizando-se dados de RMN a partir de um algorítmo genético de múltiplos mínimos

• Ano de defesa: 2009
• Autor(a) principal: Linden, Marx Gomes Van Der
• Orientador(a): Dardenne, Laurent Emmanuel
• Banca de defesa: Almeida, Fábio Ceneviva Lacerda , Bish, Paulo Mascarello
• Tipo de documento: Dissertação
• 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: Proteínas-Estrutura; Algorítmos genéticos; Ressonância Magnética Nuclear
• Palavras-chave em Inglês: Protein Structures; Genetic Algorithm; Nuclear Resonance Spectroscopy
• Área do conhecimento CNPq: CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO
• Link de acesso: https://tede.lncc.br/handle/tede/106

Proteins are biological macromolecules comprised of amino acid polymers that play a wide range of biological roles involved in every process of living organisms. Together with X-ray diffraction, Nuclear Resonance Spectroscopy (NMR) is one of the two main experimental techniques that are capable of delivering atomic-level resolution of protein structures. Prediction of proteic structures using experimental information from NMR experiments is a global optimization problem with restraints. GAPF is a computer program that uses a Genetic Algorithm (GA) developed for ab initio prediction -- the determination of the structure of a protein from its amino acid sequence only -- using a multiple minima approach and a fitness function derived from a classic molecular force field. The work presented here describes GAPF-NMR, an alternate version of GAPF that uses experimental restraints from NMR to support the search for the best protein structures that correspond to a given sequence. Five different versions of the algorithm have been developed, each with a variation on how the energy function is calculated during the course of the program run. GAPF was tested on a test set comprised of 7 proteins with known structure and it was capable of achieving a correct or approximate fold for every one of these proteins. It was noted that the versions of the algorithm that progressively increase the area of the protein used in the energy function have performed better than the other versions, and that the multiple minima approach was important to the achievement of good results. Results were compared to those obtained by GENFOLD -- to the moment, the only known alternate implementation of a GA to the problem of protein structure prediction using NMR data -- and the current version of GAPF was shown to be superior to GENFOLD for two of the three proteins that compose its test set.