Predictive Analysis of Molecular Refractivity in Quinolone Antibiotics Using Degree-Based Topological Descriptors

Quantitative Structure–Property Relationship (QSPR) analysis has become an important computational approach for understanding the relationship between molecular
structure and physicochemical properties of chemical compounds. Topological indices derived from molecular graphs provide a powerful mathematical representation of chemical
structures and have been widely used in drug discovery and cheminformatics. In this study, degree-based topological indices are employed to investigate the structural characteristics of quinolone antibiotics, a significant class of synthetic antibacterial agents widely used in the
treatment of various bacterial infections. The main objective of this research is to evaluate the
predictive capability of selected topological descriptors in estimating physicochemical properties, particularly molecular refractivity (MR), which is closely associated with molecular volume, polarizability, and intermolecular interactions.
Initially, the molecular structures of selected quinolone antibiotics are represented as molecular graphs, from which several degree-based topological indices are calculated. These descriptors capture the connectivity patterns of atoms and bonds within the molecules and serve as numerical features for predictive modeling. To establish relationships between the computed indices and molecular refractivity, different curvilinear regression models, including linear, quadratic, and cubic regression techniques, are applied. These regression models allow the exploration of both linear and nonlinear relationships between structural descriptors and physicochemical properties.
The performance and accuracy of the proposed models are evaluated using Root Mean Square Error (RMSE) as the primary statistical metric. Comparative analysis of the regression models reveals that certain topological indices exhibit strong correlations with molecular refractivity.
Among the evaluated descriptors, the harmonic index demonstrates the highest predictive capability, producing the lowest RMSE values and indicating a strong relationship between molecular topology and physicochemical behavior. The results confirm that degree-based topological indices can effectively serve as reliable predictors for molecular properties of quinolone antibiotics.