A Multivariate QSAR Study on the Anticonvulsant Activity of Acetamido-N-Benzylacetamide Derivatives. Influence of Different Molecular Descriptors

Abstract

A quantitative structure activity relationship analysis was applied to a library of 51 benzylacetamide derivatives with anticonvulsant activity. The molecular structures of 51 compounds were optimized with the Semiempirical Method PM6 (Parametric Method-6) included in the MOPAC2009 software. The optimized structures of all the examined compounds were represented by 1497 DRAGON-type descriptors. Using multiple linear regression (MLR), the influence of constitutional, topological, electronic, physicochemical and quantum molecular descriptors on the activity was investigated. The validation of models were performed through the leave-one-out cross technique in conjunction with external validation. From different types of molecular descriptors, six new QSAR models were developed. The models using 0D descriptors (constitutional descriptors) and quantum descriptors do not have high statistical quality. The combination of 2D and 3D descriptors produce a model of high predictive quality (Rcal= 0.888; Scal= 0.195; Rval= 0.814; Sval=0.431; Rloo=0.867; Sloo=0.212). Results show the important role of electronic and topologic features of molecules on their anticonvulsant activity in relative to constitutional parameters.