Modelo preditivo qsar para avaliação da atividade de inibidores de acetilcolinesterase

Abstract

Alzheimer's disease is a progressive neurodegenerative disorder associated with progressive memory loss that afflicts millions of people worldwide, and at the moment only one class of drugs is used to treat symptoms of AD, cholinesterase inhibitors (ChE). In this work, the biological activity data (IC50) of potential acetylcholinesterase inhibitors (AChE) were evaluated, being 07 derivatives of Piperidinone, 7 derivatives of cativic acid and one derived from tacrine, described in the literature. to obtain a predictive QSAR model employing multiple linear regression. The molecular descriptors of the derivatives synthesized by the authors were calculated using the PADEL software. The selection of descriptors with the greatest influence on the biological activity (IC50) of the substances was calculated using Microsoft Excel®. The predictive model was obtained using the descriptors using the BuidQSAR program. The calculated descriptors include molecular properties with topological, electronic and geometric aspects. The importance of each descriptor was attributed according to its influence on the biological activity of the molecule. After obtaining the list of molecular descriptors, it was found which of these descriptors had the highest correlation with the biological activity of the substances, those with an absolute correlation greater than 0.1805 were selected to build the predictive model and transferred to BuildQSAR. For multivariate calibration using MRL, the best QSAR model proved to be stable, robust, with low residual values and predictive for biological activity based on the substances studied, with values of q2 = 0.986, r2 = 0.996 and r2 (adjusted) = 0.992 associated with MATS6v, XlogP, ATSC8c and ZMIC1. The predictive equation presented the following coefficients: pIC50 = - 1.0107 (±0.0893) MATS6v - 0.3741 (± 0.0918) XlogP + 0.1478 (± 0.0853) ATSC8c - 0.0941 (± 0 , 0878) ZMIC1 - 0.7575 (± 0.0744) All coefficients showed statistical significance at the 95% confidence level. The correlation coefficients between the descriptors obtained were less than 0.450. The results presented showed that the predictive model obtained allowed to obtain reliable and predictive models, providing information for the synthesis of new substances based on the structure and potentially more active in the inhibition of acetylcholinesterase.