Abstract

Molecular substructures and fragment descriptors are important tools for many computational applications in medicinal chemistry and drug design. Here we briefly review the history of molecular fragmentation methods and describe the currently most widely used approaches. Then we focus on random fragmentation methods that have only recently been introduced and discuss selected applications. It is shown that the generation and mining of random fragment populations makes it possible to identify novel types of fragment descriptors and generate similarity search tools that depart from conventional design approaches. Thus, random fragmentation schemes complement and further extend established fragment methods.

Keywords: Substructures, fragment descriptors, systematic and knowledge-based fragment design, random fragmentation, molecular similarity, structure-activity relationships, database mining, computational medicinal chemistry