Predictive QSAR models development and validation for human ether-a-go-go related gene (hERG) blockers using newer tools

Abstract

In the present computational analysis, pharmacophore-based active conformer selection method was used to derive active conformers for the physicochemical descriptors calculation. The significant regression models were validated using different validation methods, which provided significant Q2 values. The distance-based approaches were also used to analyze the discriminant property of the molecules contributed in the models. The Mahalanobis distance (MD) values obtained from these studies revealed that the compounds with very high and very low acting human ether-a-go-go-related gene blockers possessed high MD values, while the predicted activity of those compounds exhibited less residual errors. The results obtained in the studies suggest that the distance-based approaches can be used to validate the quantitative structure-activity relationship models significantly. The descriptors contributed in the models explain that the flexibility of the bonds connected to the aromatic rings or non-polar region of the molecules make π–π interaction with the aromatic residues of the protein.