Desenvolvimento de metodologias de novo para predição de estruturas de proteínas

• Autor(a) principal: Oliveira, Raphael Trevizani Roque de
• Data de Publicação: 2014
• 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/196
• topic: Proteinas; Predição de estruturas; Bibliotecas de fragmentos; Biologia molecular - Simulação por computador; Protein; Prediction structures; CIÊNCIAS BIOLÓGIAS : BIOQUIMICA; CIÊNCIAS EXATAS : CIÊNCIA DA COMUTAÇÃO

De novo protein structure prediction aims to find the 3D conformation of a protein from its amino acid sequence without the use of experimental templates. One of the most successful strategies consists in assembling models from a collection of small fragments of other proteins using a search algorithm. GAPF (Genetic Algorithms for Protein Folding) is a software for ab initio protein structure prediction developed by GMMSB/LNCC which uses a multiple minima genetic algorithms (GA) to search the energy landscape. The aim of this work is to incorporate a de novo methodology to GAPF to increase its predictiveness. The main strategy implemented is based on using fragment libraries. Fragments are selected based on sequence similarity and secondary structure prediction, and were used both to assemble the individuals of the initial population, and as mutation operators. We developed a strategy to insert fragments whose length was determined using the confidence of the secondary structure prediction. Additionaly, the structures with the highest hydrophobic compactness were favoured by a new form of parental selection. The test set comprises 20 proteins distributed among mainly-α, mainly-β and α/β classes, ranging from 20 to 146 aminoacids. The de novo method presented here was able to improve the prediction for 75% of the proteins of the test set, and the improvement was considered significative for 50% of the proteins of the test set. Besides the performance improvement (i.e., smaller number of evaluations of the energy function), a greater number of individuals with better hydrophobic compactness was generated. The results of this work point to important pathways to better de novo methods and aided setting the protocol that allowed GAPF to participate in the Critical Assessment of Protein Structure Prediction - CASP 11.