Applications of double bond indices: Quantitative characterization of properties of benzenoid hydrocarbons

Topological indices (TIs) are widely used descriptors in QSAR/QSPR modeling, but conventional degree-based TIs neglect bond multiplicity, potentially limiting structural resolution. Thirteen extended double-bond-corrected TIs were derived via a Kekuúlé-structure-based edge classification and benchmarked against their conventional counterparts across 100 benzenoid hydrocarbons and 11 physicochemical properties using nested F-tests, FDR-corrected significance testing, and a permutation-based null model to isolate structural contributions from mere numerical continuity. Extended indices resolved substantially more numerical ties than conventional indices (Wilcoxon p = .0010); however, permutation benchmarking showed that this improvement reflects the continuous-valued construction rather than the specific double-bond information encoded (p ≈ 1.00). Regression analyses further showed neither index family exceeded simple size-based baselines (molecular weight, ring count) for most properties, with a marginal exception for log P. This rigorous evaluation clarifies the genuine, more limited scope of double-bond-corrected TIs and establishes a reproducible statistical framework for evaluating future topological descriptors.