Estudos SAR e QSAR-2D de derivados de N-benzoil-2hidroxibenzamidas ativos contra Plasmodium falciparum

• Ano de defesa: 2017
• Autor(a) principal: Freitas, Verlucia Amanda Machado de
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal da Paraíba; Brasil; Química; Programa de Pós-Graduação em Química; UFPB
• Palavras-chave em Português: SAR; QSAR-2D; P. falciparum; Benzamidas; 2D-QSAR; Benzamides; CNPQ::CIENCIAS EXATAS E DA TERRA::QUIMICA
• Link de acesso: https://repositorio.ufpb.br/jspui/handle/123456789/11873

Among parasitic diseases, malaria is one of the most devastating. One of the causative agents is Plasmodium falciparum, which is responsible for the highest mortality rates. Although there are some drugs employed in the treatment of malaria, issues such as parasitic resistance and low treatment efficacy make extremely necessary the design of new drugs. There is a huge commercial interest in biological activity prediction of new molecules and a largely utilized methodology are the structure-activity relationship studies. In this work, a set of 39 N-benzoil-2-hydroxybenzamide derivatives were employed and some electronic properties were calculated using DFT method with the M06-2X functional and the 6-311+g(d,p) basis set, along with physical-chemical and structural descriptors, using the on-line platform E-dragon 1.0. Analyses of structure-activity relationships (SAR) and bi-dimensional quantitative structureactivity relationships (2D-QSAR) were performed with different chemometric techniques classified in three categories: unsupervised pattern recognition (HCA and PCA), supervised pattern recognition (KNN, SIMCA and PLS-DA) and multivariate calibration (PLS and MLR) with variable selection for classification using methods of Fisher weights and GA (Genetic Algorithm), and variable selection for calibration using OPS (Ordered Predictor Selection) and GA. In HCA and PCA, characteristic clusters and the separation of active and inactive samples were verified. In classification techniques, SIMCA and PLS-DA models have demonstrated reliability and good internal consistency with Correct Classification Rate above 90% for training and test set. The selected descriptors suggest that there are structural features which allow the separation of active and inactive compounds in the chemical space defined. For multivariate calibration, variable selection with OPS along with PLS has led to a better QSAR model proposal, which has demonstrated to be stable, robust and predictive for antimalarial activity of the class of compounds under study, with correlation coefficients q2 = 0,75, r2 = 0,81 and r2pred = 0,89. For the best model, antimalarial activity is associated to descriptors of 2D autocorrelation descriptors, Edge adjacency indices, Information Indices, WHIM, RDF and 3D-Morse descriptors, and the quantum chemical polarizability descriptor. The results indicate that the developed procedure for the set of N-benzoil-2-hydroxybenzamides allow the achievement of reliable and predictive models, providing subsides for synthesis and biological evaluation of new compounds with structural features similar to the ones studied here and potentially more active against P. falciparum.