Predicting Physicochemical Properties of Antibiotic Drugs via Neighborhood Degree-Based Topological Indices and QSPR Models

Quantitative structure–property relationship (QSPR) models generate a mathematical formalism describing the molecular structure with the help of physical and chemical properties of compounds. In the present work, we are reporting an investigation on a set of neighborhood degree-based topological indices for QSPR studies of some antibiotic drugs such as amoxicillin, ampicillin, ciprofloxacin, gemifloxacin, norfloxacin, oxytetracycline, sulfadimidine, sulfamethoxazole, sulfamethoxypyridazine, sulfanilamide, tetracycline, and thiram. For each drug there exists a modeled molecular graph, and some neighborhood degree-based descriptors are calculated. The evaluated topological descriptors are used in QSPR models based on linear regression and logarithmic regression for the relationship between the molecular structure and physicochemical properties. The statistical study confirms that the neighborhood degree-based descriptors have a robust predictive power, indicating an efficient molecular descriptor. We find that descriptors at the level of neighborhood degree are highly correlated with a variety of physicochemical properties of antibiotic molecules. The results underscore the mathematical significance and utility of these indexes in QSPR modeling and shed more light on the structural characterization and theoretical evaluation of biologically active molecules. This study is of use for gaining some insights regarding the structure–property relationship of the antibiotic drugs and justifies using a graph-theoretical descriptor approach in drug property predictions.