In silico ADME modeling: QSPR models for the binding of β- lactams to human serum proteins using genetic algorithms

Abstract

Quantitative structure-property relationship (QSPR) models based on in vitro serum proteins binding data of 113 diverse drugs and drug-like compounds are reported. For this purpose, two Genetic Algorithm (GA) based approaches GA1 and GA2 along with multiple linear regression (MLR) are employed to exhaustively search and to select multivariate linear equations, starting from a large pool of molecular descriptors (molecular properties or variables). The reported QSPR models are based on combinations of 5 and 6 molecular properties calculated from the 2D chemical structures. Internal (leave-one-out) and external validation tests have demonstrated that these models have excellent predictive power and can be applied to the design new β-lactams class of antibiotics. As the models reported herein, are based on computed properties, they appear as valuable virtual screening tools, where selection and prioritisation of candidates is required.