Partial Molecular Alignment via Local Structure Analysis

Molecular alignment remains as one of the most problematic aspects of molecular design. A technique is introduced that facilitates the alignment of a range of structures that could not be handled easily using existing alignment procedures. The flexibility of the method is illustrated with a series of test sets. First, an alignment is performed on a series of molecules from a typical 3D-quantitative structure- activity relationship data set. The results of this test show the technique to outperform many existing alignment methodologies based upon the optimization of molecular similarity or molecular overlaps. This test set is then extended to consider the alignment of more structurally diverse inhibitors of HIV-1 reverse transcriptase and HIV-1 protease. Finally, in the most challenging test, a large protein-based inhibitor is matched with a smallmolecule mimic. It is believed that the existence of such a versatile alignment technique will prove invaluable in the fields of molecular design and chemical information handling.