Computational Studies on Maraviroc Derivatives for HIV Treatment: QSAR and Molecular Docking Approaches

This research utilized four different methods to investigate the structure-activity relationships of 25 derivatives of Maraviroc. The combination of Genetic Algorithms-Artificial Neural Networks (GA–ANN) and Multiple Linear Regression-Imperialist Competitive Algorithm (MLR-ICA) demonstrated superior performance among both linear and nonlinear methods. The study identified specific descriptors, such as atomic van der Waals volumes, polarizability, and atomic masses, as significant in the Genetic Algorithms–Artificial Neural Networks (GA-ANN) method for biological activity assessment. In terms of lipophilicity, descriptors related to Verhaar Algae base-line toxicity and polarizability were highlighted in the Multiple Linear Regression-Imperialist Competitive Algorithm (MLR-ICA) method. Molecular docking analysis revealed that Maraviroc derivative number 22 with 5UIW receptor exhibited the lowest affinity but the highest number of hydrogen bonds. The Monte Carlo technique, utilizing CORAL software, pinpointed essential molecular characteristics linked to both biological activity (–logIC50) and lipophilicity (XLOGP). These features encompassed the existence of cyclic rings with branching, sp2 carbon linked to a ring, exclusive presence of double bonds, Nitrogen attachment to cyclic rings, Nitrogen presence in double bonds, Fluorine atom connection to branching, presence of branching, and Nitrogen atom linkage to a ring. The research found that the combined use of GA-ANN, MLR-ICA, Monte Carlo method, and molecular docking can enhance understanding of the relationship between physico-chemical descriptors and drug mechanisms, aiding in the design of new drugs.