Solvation-weighted connectivity indices for GC retention prediction of polycyclic aromatic sulphur heterocycles

Predicting the gas chromatographic retention indices (RIs) of sulphur-containing polycyclic aromatic hydrocarbons (S-PAHs) is challenging due to positional isomerism, which introduces subtle structural variations that are not well captured by conventional physicochemical descriptors or simple partitioning models. In this study, a quantitative structure-retention relationship (QSRR) model was developed to accurately predict the RIs of S-PAHs using three molecular descriptors calculated with AlvaDesc software. Among the evaluated variables, solvation-weighted connectivity indices (X3sol, X4sol, and X5sol) emerged as the most influential descriptors governing chromatographic behaviour and selectivity. The model was built using a training set of 35 compounds and externally validated with an independent test set of 13 compounds. It demonstrated excellent predictive performance, with an adjusted coefficient of determination (R²adj.) of approximately 0.99 for both datasets. Residual analysis confirmed the robustness of the model, showing deviations mostly within ±5 RI units and no systematic error. Overall, solvation-weighted connectivity indices effectively capture subtle topological differences among positional isomers, providing a robust, interpretable, and reliable framework for predicting and understanding the retention behaviour of S-PAHs.