Comprehensive 4D-QSAR Analysis for Predicting the Biological Activity of Quinazoline: Integrating Electron Conformational and Genetic Algorithm Approaches through a Unified Framework to Uncover Essential Pharmacophoric Features

Abstract

Abstract To unravel the intricate connection between the molecular structure and bioactivity within a series of quinazolines, our investigation adopted a synergistic methodology that employed a genetic algorithm (GA) in tandem with four-dimensional quantitative structure-activity relationship (4D-QSAR) modeling. Rather than relying on a singular conformation, our model construction represented each compound with a set of conformers. The geometric and electronic structure attributes for every atom and bond in each molecule were computed and organized into an electron–conformational matrix of contiguity (ECMC). The electron conformational submatrix of activity (ECSA) was derived through a comparative analysis of these matrices. For the series of quinazolines, we developed a pharmacophore model based on chemical properties utilizing the EMRE software package. Employing a genetic algorithm, we identified crucial variables to predict theoretical activity. The training set, consisting of 41 compounds, was used to develop 4D-QSAR models, and their predictive capacity was evaluated by including an additional 20 compounds in the test set. The model, incorporating the top twelve parameters, exhibited satisfactory performance. To further scrutinize the contribution of each descriptor to biological activity within the EC–GA model, the E statistics technique was applied.