Premium
Use of quantitative structure–enantioselective retention relationship for the liquid chromatography chiral separation prediction of the series of pyrrolidin‐2‐one compounds
Author(s) -
Rasulev Bakhtiyor,
Turabekova Malakhat,
Gorska Magdalena,
Kulig Katarzyna,
Bielejewska Anna,
Lipkowski Janusz,
Leszczynski Jerzy
Publication year - 2012
Publication title -
chirality
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.43
H-Index - 77
eISSN - 1520-636X
pISSN - 0899-0042
DOI - 10.1002/chir.21028
Subject(s) - chemistry , heteroatom , enantiomer , enantioselective synthesis , homo/lumo , molecular descriptor , chirality (physics) , quantitative structure–activity relationship , linear regression , computational chemistry , retention time , chromatography , molecule , stereochemistry , organic chemistry , ring (chemistry) , machine learning , nambu–jona lasinio model , chiral symmetry breaking , quantum mechanics , computer science , quark , catalysis , physics
In this work, the enantioseparation of 15 structurally similar chiral solutes is studied, and analysis of the retention factors is performed using a genetic algorithm and multiple linear regression analysis technique. The present quantitative structure–enantioselective retention relationship model generated for retention factors' data has confirmed the importance of a number of descriptors altering the retention behavior and enantioselectivity of the studied compounds. Thus, fragment‐based descriptor PSA, which encodes polar surface area, has confirmed the crucial role of heteroatoms in the retention behavior exhibited by pyrroliddin‐2‐ones. The presence of E LUMO descriptor, which represents a quantum‐chemical property, has indicated the role of charge transfer interactions between the chiral stationary phase and enantiomers to retention factors, showing that lowest unoccupied molecular orbital energy is significantly different between enantiomers. The developed model exhibits a very good performance characterized by following statistical parameters: r 2 = 0.93 for training set and r 2 = 0.99 for the validation set. The selected three‐variable model displays high predictive ability, catching the main factors affecting the enantioselectivity of studied chiral compounds, and therefore can be used for prediction of retention factors of other chiral compounds of similar structure to improve the separation process and so on. Chirality, 2011. © 2011 Wiley Periodicals, Inc.