3D quantitative structure activity relationships with CoRSA. Comparative receptor surface analysis. Application to calcium channel agonists

The in vitro or in vivo interaction between chemical compounds and their biological targets (transporters, receptors, ion channels, enzymes) can be efficiently predicted with the 3D QSAR models introduced in recent years. In this paper we describe CoRSA (comparative receptor surface analysis), a novel 3D QSAR algorithm that can be applied to compute structure-activity equations whenever the structure of the biological target is not known. Using the common steric and electrostatic features of the most active members of a series of compounds, CoRSA generates a virtual receptor model, represented as points on a surface complementary to the van der Waals surface of the set of compounds. The structural descriptors of the model are represented by the total interaction energies between each surface point of the virtual receptor and all atoms in a molecule. These descriptors are used in a partial least squares (PLS) data analysis to generate a structure-activity model. A highly significant CoRSA model was obtained for a set of compounds that act as calcium channel agonists for the guinea pig left atrium assay.