Desenvolvimento e implementação de um modelo Coarse-Grained para predição de estruturas de proteínas

• Autor(a) principal: Goliatt, Priscila Vanessa Zabala Capriles
• Data de Publicação: 2011
• Tipo de documento: Tese
• Idioma: por
• Título da fonte: Biblioteca Digital de Teses e Dissertações do LNCC
• Texto Completo: https://tede.lncc.br/handle/tede/157
• topic: Estrutura molecular; Algorítmos genéticos; Genetic algorithms; Molecular structure; CNPQ::CIENCIAS BIOLOGICAS::BIOQUIMICA::BIOLOGIA MOLECULAR

The knowledge of the three-dimensional (3D) conformation of the native structure of proteins has assisted biotechnological researches as those focused on protein design and structure based drug design. As an alternative to the techniques of experimental prediction and theoretical prediction via comparative modeling, the theoretical prediction via ab initio technique has been increasingly exploited. In the ab initio technique, the 3D protein structure prediction (PSP) is performed based only on its amino acid sequence and in their physicochemical properties. However, to obtain solutions that represent native structures, methods of ab initio PSP require a large number of function evaluations which limits their use in studies of more complex proteins. In the CG model, the representation of each amino acid is reduced to a set of few interaction sites, significantly reducing the number of degrees of freedom to be treated, and thus reducing the computational cost needed to evaluate each structure during the optimization process. In this work, we developed and implemented a CG model applied to the PSP problem, including the ability to simulate the presence of disulfide bonds. This model was adapted to the GAPF program (Genetic Algorithm for Protein Folding), previously developed by the "Grupo de Modelagem Molecular de Sistemas Biológicos" (GMMSB/ LNCC) and originally implemented for the PSP via ab initio technique using an all-atom model representation. The CG model has been tested on a set of 36 structures with sizes ranging from 18 to 106 amino acid residues, and covering the different classes of proteins (i.e. mainly alpha, alpha+beta and mainly beta). The results showed that the CG model proposed promoted improvements in predictive ability of GAPF program, now being able to predict β-sheet structures, and a decrease of approximately 80% of computational time.