A simple QSPR model to predict aqueous solubility of drugs

Abstract

Aqueous solubility of a drug/drug candidate is essential data in drug discovery, and an in silico method for predicting the aqueous solubility of drug candidates provides a valuable tool to speed up the process of drug discovery and development. This paper describes a simple quantitative structure property relationship (QSPR) model for predicting the aqueous solubility of drugs which is validated by cross-validation methods. A data set of 220 drug or drug like molecules as a train set was employed and the accuracy of the proposed QSPR model was compared with those of the general solubility equation (GSE) and the linear solvation energy relationship (LSER). Also, a test set containing the aqueous solubility of 75 official drugs which are structurally and physico-chemically diverse, was proposed to compare the accuracy of the aqueous solubility prediction models as a reference data set. The developed model is: log S(w) = -1.120E - 0.599 ClogP, in which is the molar aqueous solubility of a drug, E is the excess molar refraction and ClogP is the computed logarithm of partition coefficient of drug. The E and ClogP values for a drug candidate could be computed using Pharma-Algorithms software. Average absolute error (AAE) and mean percentage deviation (MPD) were used as comparison criteria. The proposed QSPR provided better AAE and MPD for solubility prediction in comparison with GSE and LSER models.