Automated generation of pharmacophore type constraints to improve FlexX docking

CINF 74

Andrea Volkamer, Thomas Lengauer, lengauer@mpi-inf.mpg.de, and Andreas Kämper, akaemper@mpi-inf.mpg.de. Computational Biology and Applied Algorithmics, Max Planck Institute for Informatics, Stuhlsatzenhausweg 85, D-66123 Saarbrücken, Germany
The use of pharmacophore constraints is an established technique for improving docking results. Usually the pharmacophore has to be specified manually. In this work we present a fully automated technique that incorporates information on the buriedness of the binding pocket and structure-based pharmacophore features into the docking engine of FlexX. Key interaction points in the active site are calculated with a GRID-based energy function. Those points that pass several newly developed filters are merged to a small number of pharmacophore features. The automatically generated pharmacophores agree well with manually derived results. The performance of the method has been validated on several difficult docking tasks as well as on virtual screening scenarios using FlexX-Pharm, the pharmacophore module of FlexX. Docking results are improved in 95 % of the test cases. In general, the enrichments in virtual screening runs are higher and the compute-times are smaller than in the respective unconstrained screenings.