The calculations done on World Community Grid will predict which small molecule compounds, out of the millions contained in a library database, should be tested for their ability to inhibit the flavivirus protease. This is a major step towards the ultimate goal of discovering new drugs to stop flavivirus infections.
Phase 1 of this project predicted how each small drug molecule might bind to the active site of the viral protease. This phase also produced preliminary "energies" that coarsely rank the strength of the intermolecular interactions between the compounds and viral protease.
Phase 2 will accurately predict free energies of binding between each drug compound and the viral protease. This calculation utilizes the binding orientations calculated in Phase 1. Due to computation time required for each free energy of binding calculation, only compounds with "good" scores from Phase 1 will be selected for Phase 2 calculations.
As an analogy, Phase 1 will tell us how two people might hold hands, whereas Phase 2 will tell us whether or not they want to hold hands.
Phase 2 of our project is designed to reduce the number of Phase 1 false positives (i.e., dead ends) that are tested in our laboratory. Phase 2 will take several thousand Phase 1 hits, run each hit through computationally demanding free energy calculations, and remove many of the false positives from the hit list. Phase 2 is expected to produce an updated list of hits that contains ~80% true positives. Testing Phase 2 hits in the laboratory will be much more productive, efficient, and rewarding than testing Phase 1 hits. For instance, to find 25 small molecules that stop dengue virus replication in the laboratory, we would need to synthesize and test either 250-500 Phase 1 hits or ~30 Phase 2 hits.