Modelling for antimicrobial activities of ionic liquids towards Escherichia coli, Staphylococcus aureus and Candida albicans using linear free energy relationship descriptors

To predict antimicrobial activities i.e., minimal inhibitory concentration (MIC) and minimal biocidal concentration (MBC) for ionic liquids (ILs) against Escherichia coli, Staphylococcus aureus and Candida albicans, six quantitative structure-activity relationship (QSAR) models were developed using linear free energy relationship (LFER) descriptors calculated by density functional theory and conductor screening model. The LFER descriptors are excess molar refraction, dipolarity/polarizability, H-bonding acidity, H-bonding basicity, McGowan volume, cationic interaction, and anionic interaction. By excluding some descriptors with ignorable contributions to training set, components of the QSAR models were simplified. Their estimated predictabilities were in R2 = 0.900, standard error (SE; in log unit of μM) = 0.430 for log 1/MIC of E. coli, R2 = 0.934, SE = 0.370 for log 1/MBC of E. coli, R2 = 0.910, SE = 0.470 for log 1/MIC of S. aureus, R2 = 0.947, SE = 0.350 for log 1/MBC of S. aureus, R2 = 0.892, SE = 0.362 for log 1/MIC of C. albicans and R2 = 0.803, SE = 0.233 for log 1/MBC of C. albicans. Then, except for log 1/MBC of C. albicans due to lack of data points, the models were validated by comparing between observed and calculated values of test set; its checked correlations were all within R2 of 0.921.