QSRR modeling for GC-MS retention prediction of emerging Xenobiotic contaminants: a robust in silico tool for environmental screening and risk assessment

The occurrence of emerging xenobiotic contaminants in the environment poses significant risks to ecosystems and human health, necessitating rapid and reliable analytical methods for their monitoring. Quantitative structure–retention relationship (QSRR) analysis has emerged as a powerful in silico technique for linking molecular structure to chromatographic behavior, enabling the prediction of analytical parameters without reference standards. In this work, we developed a robust QSRR model for predicting the gas chromatography–mass spectrometry (GC-MS) retention times (RTs) of a diverse set of xenobiotic compounds, including potential emerging contaminants. Molecular geometries were optimized using the PM7 semi-empirical method, from which a wide range of theoretical descriptors were computed. A Genetic Algorithm (GA) was employed for feature selection, yielding an optimal five-descriptor combination (DBI, R8i, TPSA_efficiency, Hy, and SM1B(p)). The model was rigorously validated using both internal and external validation strategies. Internal validation demonstrated excellent statistical performance (R² = 0.88, Q²_LOO = 0.877, s = 0.11, RMSE_tr = 0.11). External validation confirmed the model's strong predictive power (Q²_F1 = 0.812, Q²_F2 = 0.810, Q²_F3 = 0.837, CCC_ext = 0.89, RMSE_ext = 0.138). These results underscore the robustness and practical utility of the developed QSRR model as a reliable tool for estimating the GC-MS retention times of emerging xenobiotics. This approach facilitates their rapid screening and identification in complex environmental matrices, thereby supporting exposure assessment and contributing to the broader understanding of their environmental risks and potential health effects.