A simple, knowledge-based scoring function has been developed that is based on frequent geometric and chemical patterns of interacting atoms found at the interfaces of x-ray crystalized protein-ligand complexes. Test protein-ligand complexes are scored based on the atomic interaction patterns found at their interface that match these “classical” frequent patterns (FP). The FP scoring function has been validated for its ability to rank the pose closest to native pose of a ligand in the X-ray crystal structure of the protein-ligand complexes among other non-native poses (decoys) produced by computational docking. Our solution has shown that this novel FP pose scoring function affords higher accuracy of ranking the pose closest to native as compared to seven other commonly used commercially available scoring functions.
• Experimentally validated to be a robust identifier of protein-ligand interactions
• Simple FP scoring function based on chemical and geometrical similarities in the complex-specific interaction patterns
• FP scoring function capable of identifying the correct binding modes for protein-ligand complexes