QSRR-based insights into the chromatographic behaviour of phenolic derivatives on biomimetic stationary phases

A quantitative structure-retention relationship (QSRR) approach was employed to investigate the chromatographic behaviour of phenolic derivatives on biomimetic stationary phases. The study aimed to elucidate the molecular determinants governing lipophilicity, membrane affinity, and protein binding using two complementary chemometric strategies: genetic algorithm-partial least squares (GA-PLS) and canonical correlation analysis (CCA). GA-PLS models, developed separately for each chromatographic endpoint, exhibited excellent predictive performance ((Formula presented), (Formula presented) ) and revealed phase-specific determinants related to molecular topology, charge distribution, and lipophilicity. The CCA model, designed to integrate all chromatographic systems simultaneously, identified shared structural patterns driving retention across phases. Although slightly less predictive, it achieved high CCA (R2[jls-end-space/]>0.92) and maintained satisfactory external validity while offering superior interpretability. Mechanistic interpretation indicated that long-range steric and electronic complementarity, together with aromaticity and polarizability, governs the chromatographic behaviour of phenolic compounds across biomimetic systems. The findings demonstrate that GA-PLS and CCA provide complementary insights, with the former excelling in system-specific prediction and the latter enabling a unified understanding of cross-system molecular interactions. This study represents the first application of CCA to the simultaneous modelling of multiple biomimetic chromatographic systems, expanding the methodological scope of QSRR in bioanalytical research.