Topological indices and QSPR analysis of drug molecules from different therapeutic classes

Abstract Typically, Quantitative Structure–Property Relationship (QSPR) models are created for compounds related to a particular therapeutic use and might be restricted in scope. To explore the wider use of neighbourhood connectivity descriptors, ten drug molecules, encompassing a wide range of chemical and therapeutic classes, were chosen: anti-inflammatory, anti-bacterial, anti-viral, anti-hypertensive, anesthetic, anti-depressant and neuromuscular agents. The molecular structures were represented as graphs, with atoms represented by the vertices and chemical bonds represented by the edges, and various neighbourhood degree-based topological indices were calculated. Linear, quadratic and cubic QSPR regression models were used to correlate these descriptors with nine experimentally determined physicochemical properties such as boiling point, density, enthalpy of vaporization, flash point, refractive index, molar refractivity, polarizability, surface tension and molar volume. The coefficient of determination ( $$\textrm{R}^{\circ }$$ ) was used to evaluate the model performance. The results show that neighbourhood-based descriptors are able to capture molecular structural information and make reliable prediction of properties on structurally diverse drug molecules. The results provide good justification to use these descriptors in general for chemical graph theory and in QSPR studies without any particular disease or therapeutic indication.