QSAR Regression Models for Predicting the Activity of Inhibitors of Staphylococcus Epidermidis

Abstract

Staphylococcus epidermidis is a common symbiont bacteria, and these are common causes of nosocomial infections. The QSAR model for the identification of inhibitors against Staphylococcus epidermidis is developed. A database of around 600 compounds with known inhibitory zone values was collected. Descriptors representing structural and functional properties were calculated using Alvadesc v.1.02. This database was curated using feature selection and extraction procedures. The model was developed using four different machine learning algorithms on a training set called multiple linear regression, support vector regression, random forest regression, and random tree regression. The model performance has been assessed by traditional metrics including Pearson's correlation coefficient, coefficient of determination, mean absolute error, and root mean square deviation. The author also has performed external validation, cross-validation, and y-randomization test. Models with reasonably good performance were built and can be used for the virtual screening of inhibitors.