Novel consensus quantitative structure-retention relationship method in prediction of pesticides retention time in nano-LC

In this study, quantitative structure-retention relationship (QSRR) methodology employed for modeling of the retention times of 16 banned pesticides in nano-liquid chromatography (nano-LC) column. Genetic algorithm-multiple linear regression (GA-MLR) method employed for developing global and consensus QSRR models. The best global GA-MLR model was established by adjusting GA parameters. Three descriptors of SpMax2_Bhp, Mor31u and, MATS6c appeared in this model. Consensus QSRR models developed as an average consensus model (ACM) and weighted consensus model (WCM) by a combination of a subset of the GA-MLR models. Comparison of statistical parameters of developed models indicated that an ACM which is combining of the best global QSRR model with four-descriptor sub-model can be selected as the best consensus QSRR model. CrippenLogP, RDF070m, Lop, and HASA1 descriptors appeared in four-descriptor sub-model. In ACM, the square of correlation coefficients (R2) was 0.973 and 0.939, and the SE was 0.49 and 0.40, for the training and test sets, respectively. The ACM was assessed by leave one out cross-validation ("Q2 cv" = 0.935) as well as internal validation. Descriptors which appeared in this model suggest electrostatic, steric and hydrophobic interactions play the main role in the chromatographic retention of studied pesticides in nano-LC conditions.