In silico renal clearance model using classical Volsurf approach

Abstract

A data set of 130 diverse compounds containing both central nervous system (CNS) and non-CNS drugs was used to generate a renal clearance model using a classical Volsurf approach. Percentage renal clearance data was used as a biological input. The score plots obtained from principal component analysis and partial least-squares (PLS) analysis clearly separated high-clearance compounds from low-clearance compounds. PLS models were used to predict the renal clearance of the data set. Categorical statistical methods such as SIMCA and recursive partitioning techniques were used for classifying the compounds into low-and high-clearance categories. PLS coefficient plots, Volsurf descriptor profiles, 3D Grid maps, and RP decision trees were used to explain the important descriptors separating low and high renal clearance compounds. For comparative purposes, topological descriptors such as Molconn-Z were also examined. All the models were validated by an external test set of 20 compounds. These models can be used as efficient tools in the classification and prediction of the renal clearance of unknown compounds, the knowledge of which is helpful in understanding their bioavailability behavior.