Diseño QSAR para inhibidores de acetilcolinesterasa

Abstract

This paper shows the results of QSAR quantitative analysis about the correlation between the structure of a set of molecules and their acetylcholinesterase inhibitor activity, a substance that has been associated with Alzheimer's disease. For analyze the obtained data, some multivariate classification methods have been used, for organize different molecules according to their structure, and thus establishing some guidelines to propose a model that allows predicting the biological activities of molecules that are different to the ones used in the training set. The best molecular descriptors for three different kinds of molecules were chosen. The first one of them is formed by molecules with a diversity of structures, the second one contains pyridine groups, and the last one consists in structures containing carbamate groups. The reliability of the method is confirmed through the KNN (k-nearest neighbors) method, obtaining a classification error of 5% leave-more-out.