Submitted by Ricardo Cedraz Duque Moliterno (ricardo.moliterno@uefs.br) on 2015-07-27T21:18:33Z No. of bitstreams: 1 Dissertac?o Jos? Jorge Silva Junior.pdf: 3038189 bytes, checksum: be43dcbd4f764e08ab18fd95a388e950 (MD5) / Made available in DSpace on 2015-07-27T21:18:33Z (GMT). No. of bitstreams: 1 Dissertac?o Jos? Jorge Silva Junior.pdf: 3038189 bytes, checksum: be43dcbd4f764e08ab18fd95a388e950 (MD5) Previous issue date: 2014-07-31 / Coordena??o de Aperfei?oamento de Pessoal de N?vel Superior - CAPES / Pests are responsible for high losses in cocoa production in Brazil and other countries, among them the witches' broom (WB) is one of the most important and destructive to the cocoa, even causing losses of up to 95% of production. This plague has spread very easily in the state of Bahia due to environmental conditions that provided the spread of WB, caused by the fungus Moniliophthora pernicious. Several chemical compounds have been tested in order to prevent or eradicate WB, however it has not showed good results, so the present study aimed to perform in silico assays were obtained in order to identify inhibitors of UDP-N-acetylglucosamine pyrophosphorylase (UNAcP) of M. perniciosa. For achieve this goal, computational methods have been employed in the search for new inhibitors for UNAcP where different stages of research and evaluation were performed. The initial stage of virtual screening consisted of the choice of the scoring function, thus the following scoring functions were evaluated: Broyden Fletcher Goldfarb Shanno (BFGS) present in AutoDock VINA 1.1.2; Grid and Grid Score Score+Hawkins GB/SA both present in DOCK 6.5 and calculate the consensus score. The results were analyzed by calculating Enrichment Factor (EF) analysis of the ROC curve and its respective Area under an ROC curve (AUC). The Grid Score presented EF(5)=7.85. The ROC curve analysis allowed us to observe that the Grid Score function can identify almost 40% of active molecules with less than 10% of the database (false positive and active molecules), AUC analysis demonstrated that the Grid Score has greater accuracy (AUC=0.87). Thus, these results showed what the Grid Score was the best scoring function for this system. A database composed of molecules derived from natural sources was also used. The top ten results of virtual screening of the DOCK6.5 underwent online platform ChemGPS-NP, for the calculation of chemical descriptors. Thus, the molecules were re-categorized, based on the values of the Grid Score DOCK6.5 and chemical descriptors ChemGPS-NP. The results indicate the ZINC68592326 molecule his the best score and the analysis indicates that this has hydrophobic interactions with Ala380, Gln113, Gly112, Gly381, Ser168, Arg383, Pro221 and hydrogen bond interaction (3.32?) with Asn224. The virtual screening database of molecules derived from natural products research allowed with a universe of structures with very different characteristics. It was possible to obtain molecules with great structural diversity between the top ranking, however, also found very similar to the reference molecules. The use of chemometric methods is considered very useful and allows a systematic and consistent choice of structures, mainly by taking into account chemical descriptors and molecular characteristics, allowing for a more detailed evaluation. / Pragas s?o respons?veis por elevadas perdas na produ??o de cacau no Brasil e no mundo, dentre elas a vassoura-de-bruxa (VB) ? uma das mais importantes e destrutivas para o cacaueiro, chegando a causar perdas de at? 95% da produ??o. Essa praga disseminou-se muito facilmente no estado da Bahia devido a condi??es ambientais que proporcionaram a propaga??o da VB, causada pelo fungo Moniliophthora perniciosa. Diversos compostos qu?micos v?m sendo testados com o objetivo de prevenir ou erradicar a VB, por?m n?o foram obtidos bons resultados, portanto, o presente trabalho teve como principal objetivo realizar ensaios in silico, a fim de identificar inibidores da UDP-N-acetilglicosamina pirofosforilase (UNAcP) do M. perniciosa. Para tanto, foram empregados m?todos computacionais na busca de novos inibidores para a UNAcP, onde foram realizadas diferentes etapas de busca e avalia??o. A etapa inicial da triagem virtual consistiu na escolha da fun??o de pontua??o, assim, foram avaliadas as seguintes fun??es de pontua??o: Broyden?Fletcher?Goldfarb?Shanno (BFGS) presente no AUTODOCK VINA 1.1.2; Grid Score e Grid Score+Hawkins GB/SA ambos presentes no DOCK 6.5 e o c?lculo do escore de consenso. Os resultados foram analisados atrav?s do c?lculo de Fator de Enriquecimento (FE), analise da curva ROC e sua respectiva ?rea Sobre a Curva (AUC). O Grid Score apresentou FE(5)=7,85. A an?lise da curva ROC permitiu observar que a fun??o Grid Score consegue identificar quase 40% das mol?culas ativas com menos de 10% do banco de dados (mol?culas ativas e falso positivos), a an?lise da AUC demonstrou que o Grid Score tem maior exatid?o (AUC=0,87). Assim os resultados da avalia??o apontaram o Grid Score como melhor fun??o de pontua??o para esse sistema. Foi utilizado um banco de dados composto por mol?culas oriundas de fontes naturais. Os dez melhores resultados da triagem virtual feita no DOCK6.5 foram submetidos ? plataforma on line ChemGPS-NP, para o c?lculo dos descritores qu?micos. Assim, as mol?culas foram recategorizadas, baseando-se nos valores do Grid Score do DOCK6.5 e descritores qu?micos do ChemGPS-NP. Os resultados apontaram a mol?cula ZINC68592326 com a melhor pontua??o, a an?lise das intera??es intermoleculares indica que est? mol?cula apresenta intera??es hidrof?bicas com os res?duos Ala380, Gln113, Gli112, Gli381, Ser168, Arg383, Pro221 e liga??o de hidrog?nio do tipo aceptora com dist?ncia de 3,32? com o res?duo Asn224. A triagem virtual em banco de dados de mol?culas oriundas de produtos naturais permitiu a investiga??o com um universo de estruturas com caracter?sticas muito diversas. Foi poss?vel obter mol?culas com grande diversidade estrutural entre os primeiros do ranking, por?m, foram encontradas tamb?m mol?culas muito similares ?s mol?culas de refer?ncia. A utiliza??o de m?todos quimiom?tricos, ? considerada muito ?til e permitem uma escolha sistem?tica e consistente das estruturas, principalmente por levar em considera??o, descritores qu?micos e caracter?sticas moleculares, permitindo uma avalia??o mais criteriosa.