Searching substructures in fragment spaces

Introduction Fragment spaces (FSs) are an elegant way to model a large or even infinite number of chemical compounds and their synthetic accessibility. A FS consists of molecular fragments and a set of rules defining how fragments can be combined to products. In virtual screening experiments, FSs might include products with undesired functional groups or inadequate central building blocks. The recognition of such products, especially when they span over multiple fragments, would require their explicit construction from the FS. Due to the generally huge number of possible products in an FS, the complete enumeration is undesired or even impossible. Therefore, algorithms that perform substructure search in FSs must be able to process fragments and joining rules rather than complete molecules. Even though some algorithms that work in FSs exist [1,2], a method that excludes undesired products via substructure definition from a FS is still missing.