Exploring anticancer drug structures through vertex based resolving parameters

The generation of molecular descriptors with structural information remains a major challenge in quantitative structure–property relationship (QSPR) studies. We introduce a new class of vertex-based topological indices based on the idea of resolving degrees stemming from metric dimension theory in chemical graph theory. In contrast to classical degree-based descriptors, the introduced indices consider vertex resolvability information, thus increasing their capacity to separate structurally similar molecular graphs. These descriptors were calculated for several anticancer drugs and correlated with spatially significant physicochemical properties, namely, boiling point, molar volume, enthalpy of vaporization, flash point, and molar refractivity. The predictive performance of the linear and cubic regression models was considered. The analysis revealed statistically significant and robust correlations between the proposed indices and all the studied properties. Although linear models present interpretable relationships, unlike cubic regression models that prioritize accuracy over interpretability, the underlying nonlinear structural effects are better represented in cubic regressions, which can consistently achieve the best accuracy. The results indicate that resolving-degree-based descriptors are more predictive of structure than their qualifying counterparts, making these new topological metrics promising candidates for the QSPR modeling framework as well as in computational drug design and cheminformatics.