Multi-level structure-based pharmacophore modelling of caspase-3-non-peptide complexes: Extracting essential pharmacophore features and its application to virtual screening.

Enormous caspase-3-non-peptide crystal structures have been developed to study the structural basis of caspase-3 enzyme inhibition using active site directed small molecular design. These complexes have not been explored thoroughly to decipher the essential non-covalent interactions made by crystal ligands. We present here a multi-level analysis of these caspase-3 complexes using structure-based pharmacophore approach wherein numerous candidate pharmacophore hypotheses were assessed for its ability to cover available caspase-3 small molecular inhibitor dataset. The reliability of the resultant pharmacophores was evaluated using three different validation sets comprising focussed caspase-3 inhibitors, focussed + random decoys, and focussed + structurally similar random decoys and its performance was measured by the Güner-Henry (GH) scoring and enrichment statistics. Furthermore, the effect on excluded volumes toward caspase-3 inhibitors mapping was investigated by an iterative deletion in the structure-based models and created optimal structure-based pharmacophore models to enable effective design of caspase-3 small molecular inhibitor design.