Exploratory QSPR analysis of physicochemical properties of selected antifungal agents using topological indices

The structural-property relationships are key in the study of the physicochemical behavior of pharmaceuticals. The degree-based topological indices available through molecular graph theory are used to investigate the structure-property analysis of selected antimycotic agents in a quantitative way in this work. The molecular graphs of ten antimycotic agents were constructed, and several degree-based indices including Mᵒ₁(Υ ᵒ) , Mᵒ₂(Υ ᵒ) , Fᵒ(Υ ᵒ) , Yᵒ(Υ ᵒ) , Sᵒ(Υ ᵒ) , Gᵒ(Υ ᵒ) , Pᵒ(Υ ᵒ) , HM^α ₁(Υ ᵒ) , HM^α ₂(Υ ᵒ) , and HFᵒ(Υ ᵒ) were computed to describe their structural characteristics. To enhance descriptive reliability, this study introduces a novel integration of degree-based topological descriptors with comprehensive multivariate statistical techniques—combining linear regression, correlation profiling, and dimensionality reduction to evaluate structural-property dependencies. Statistical methods were applied to examine the connection between molecular structure and various physicochemical properties, including density, boiling point, enthalpy of vaporization, refractive index, molar refractivity, polarizability, molecular volume, surface tension, and atomic mass. The strength of the associations between topological descriptors and physicochemical parameters were analyzed using regression analysis, descriptive statistics, Pearson analysis, and Spearman analysis of correlation. The accuracy of the fitted regression models was quantitatively evaluated for all parameters and showed to be property-dependent. The range of the standard model error metrics were: MAE = 0.0234 –161.7920 and RMSE = 0.0298 –262.6852 which scale directly with the magnitude of the target physicochemical properties. Furthermore, Leave-One-Out Cross-Validation yielded cross-validated performance metrics spanning Q²LOO = -2.1417 to 0.3514, with associated errors bounded by LOOCV-MAE = 0.0320 to 191.2633 and LOOCV-RMSE = 0.0371 to 304.5087. Moreover, Principal Component Analysis (PCA) was applied to reduce data dimensionality and identify the most informative descriptors. These findings indicate that a number of topological indices have strong positive correlations with significant physicochemical characteristics, especially those of boiling point, molar refractivity, molecular volume, and atomic mass. Among descriptors that were examined, the most consistent correlations with physicochemical properties of antimycotic agents are observed in Mᵒ₁(Υ ᵒ) , Mᵒ₂(Υ ᵒ) , and HFᵒ(Υ ᵒ) , where Mᵒ₂(Υ ᵒ) achieved strong descriptive performance for polarizability ( R² = 0.989 , RMSE = 22.8565 , MAE = 17.7267 ).. The analysis of the results of PCA indicates that the first two principal components cover about 82 percent of the total variance and, therefore, the cumulative structural and physicochemical descriptors are effective in distinguishing the targeted compounds. Overall, the results indicate that degree-based topological indices are reliable descriptor of the physicochemical behavior of antimycotic agents and serve as effective tools in computational drug development and structure-property prediction.