Using data mining to improve mutation in a tool for molecular evolution

摘要: We have developed an evolutionary algorithm-based program for drug design, the molecule evoluator. This transforms known molecules into new which may improved properties relative to parent molecule. Transforming a derivative by mutation is necessary find with increased fitness. However, mutations that just randomly add and substitute atoms often result in contain undesirable chemical substructures, can therefore not be used as drugs. want knowledge about proper structures ones do not. In this research we mined large database, World Drug Index, obtain frequencies of small substructures drug-like molecules. Some our operators were subsequently modified use these frequencies. Testing on another database molecules, NCI found knowledge-based more produced existing than original mutations. suggests produce are easier synthesize compared generated using uninformed