Computational development of a protein folding predictive tool

• Autor(a) principal: Gonçalves, Luís Pedro Cardoso
• Data de Publicação: 2018
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10773/25490
• topic: Protein folding prediction; Molecular dynamics; Monte Carlo; Coarse-Grained Methods

Proteins can range from tens to thousands of amino acids, which leads to a huge number of possible conformations. Of these conformations, few have been tested by natural evolution. Understanding the way this folding is carried out naturally and all the components/forces that are present in it is still an objective today. It was then created in the last decades an area of research called protein structure prediction that aims to determine as precisely as possible the 3D structure of a protein from its sequence. It is then proposed with this thesis to create a tool that allow to predict the structure of any given protein sequence. At an early stage, will be tested three natural proteins (1CTF, 1GAB and 2LO9) from which their structures are known, but is intended that in the future any given sequence can be subsequently translated into the respective folding. The chosen proteins will be simplified by coarse-grained methods, explored in their conformational space through the Markov chain Monte Carlo method with the help of some algorithms and the most interesting candidates will be placed in molecular dynamics simulations to test their stability. The results were obtained with two different tools that are mainly distinguished by the algorithm used - parallel-tempering and ILSRR. In the first tool, the best result was an RMSD of ca. 7 Å, with the 1CTF protein, compared to the reference frame. After some modifications, which led to the creation of the second tool, an RMSD of ca. 4 Å for the same protein has achieved, and other promising results for the other two proteins