Towards selecting the virtual series of molecules using the central fragment via R-FBDD at the early stages of drug discovery.

At the early stages of drug discovery, hit compounds are often sought in order to use series they form in subsequent stages. Scaffolds are one of the most widely used approaches to organizing compounds into series. Compounds for which close analogs are hard to synthesize often turn out to be problematic for further development. Earlier, the concept of "sociability" of fragments was put forward, however that did not lead to the development of practical approaches and applied tools. We believe that even at a virtual hit finding stage it is worth focusing not on the individual molecules, but on promising series of molecules. In this paper, we explore the approaches that help to select such series of molecules. We suggest starting with searching for central fragments that can be used as scaffolds to build a series of compounds. To evaluate the quality of a fragment as a scaffold, we propose using our recently developed virtual reverse fragment based drug discovery (R-FBDD) approach. In this article, we consider a set of practically realizable metrics that we propose to use for effective selection of "sociable" and active series. The proposed metrics can be divided into two categories: energy-based and structure-based. Energy-based metrics are based on ligand efficiency. Structure-based metrics are based on the information about known compounds including a scaffold and substituents in these compounds. Finally, based on the metrics described, a pipeline to select the necessary series has been proposed.