Similarity-Based Descriptors in Drug Design

Nowadays the drug development process starts with hits obtained in HTS assays. Up to 1.000.000 compounds are biologically screened on a yes/no basis and the resulting hits are prioritised on basis of novelty, patentability, synthetic accessibility and data obtained in early ADMET (Absorption, Distribution, Metabolism, Elimination, Toxicity) profiling programs. In parallel, in silico screening approaches are gaining increasing importance. They are mainly used to select subsets of large virtual combinatorial libraries, which then should show a higher incidence for biological activity (or at least higher drug likeliness) and thus lead to increased hit rates. Due to a high rate of failures in early in vivo studies, in silico screening is also increasingly applied for early ADME profiling. One of the key problems associated with ADMET is to ensure that the drug molecules are not interacting with so called non-target proteins. Some of the key non-target proteins associated with poor bioavailability and high toxicity include the human ether-a-go-go related gene (hERG) potassium channel and the multidrug efflux pump P-glycoprotein (P-gp, ABCB1). All these proteins share a sort of promiscuity (multispecificity) in their binding interaction with ligands, which proves problematic when predicting the potential for undesirable drug-protein interactions. During our intensive studies on propafenone-type inhibitors of P-gp, we used both chemical function based pharmacophoric feature modeling as well as self organising maps for in silico screening of large compound libraries. Additionally we developed and successfully applied a new type of descriptors, the so called SIBAR- descriptors (Similarity Based SAR). These descriptors are based on calculation of similarity values between compounds under consideration and a reference set. Recent results demonstrate that the predictivity of the models obtained can be improved by tailoring the reference set to the Within the current project we will further explore the general applicability of this 3D-SIBAR approach for early ADMET-profiling, especially the prediction of P-gp and hERG interaction. This should lead to an expert system for early ADMET in silico screening of large compound libraries.