Quantitative prediction of lipase reaction in ionic liquids by QSAR using COSMO-RS molecular descriptors
•QSAR models for prediction of enzyme performance in ionic liquids.•COSMO-RS σ-profile of ionic liquids was used as numerical descriptors.•Artificial neural network based QSAR models showed good predictability (R2 > 0.91).
Linear and nonlinear quantitative structure–activity relationship (QSAR) models based on COSMO-RS screening charge density distribution (σ-profile) were developed to predict the activity and enantioselectivity of lipase from Candida antarctica lipase B (Novozym 435) and Rhizomucor miehei (Lipozyme RM-IM) in the kinetic resolution of (R,S)-1-phenylethanol with vinyl acetate in ionic liquids. The σ-profile distribution areas of ionic liquids (Sσ-profile) were used as numerical molecular descriptors to establish the QSAR models. The models were developed based on experimental data of seventeen ionic liquids (training set) and their predictability were validated with another experimental data set of five ionic liquids (testing set). The results showed that COSMO-RS σ-profile of ionic liquids could be used as excellent independent variable for prediction of enzymatic reactions in ionic liquids. The nonlinear QSAR models based on artificial neural networks methods showed better predictive ability (R2 > 0.91) than linear QSAR models based on multiple linear regression methods.
Graphical abstractFigure optionsDownload full-size imageDownload as PowerPoint slide
Journal: Biochemical Engineering Journal - Volume 87, 15 June 2014, Pages 33–40