A quantitative structure-activity relationship investigation of some NaV1.7 active compounds has been performed by repeated, random, external test set experiments employing structural descriptors (fingerprints) of signature type in combination with support vector machine (SVM) analysis using the radial basis function (RBF) kernel. The results from the investigation show remarkably stable performance from the derived in silico models in terms of statistical measures such as correlation coefficients as well as root mean squared errors (RMSEs) for the randomly selected external test sets. Also, the Bioclipse Modeling platform is utilized for introducing interpretation to the derived models.
Copyright © 2012 Elsevier Ltd. All rights reserved.