In recent years the chemical space of REadily AvailabLe (“REAL”) drug-like compounds has increased to 1010 compounds and continues to grow rapidly. However, the compound libraries used in high-throughput and virtual ligand screenings remain limited to 1-10 Mln compounds. The newly developed virtual synthone hierarchical enumeration screening (V-SYNTHES) approach enables comprehensive screening of Giga-scale space, while performing docking of only 1Mln compounds [Sadybekov et al., Nature 2022, 601, 452–459].1 The efficiency of this approach relies on the pre-docking of a Minimally Enumerated Library (MEL) of fragment-based compounds. Each MEL fragment represents a combination of one Enamine REAL reaction scaffold and one corresponding synthon. If selected, the MEL fragment is enumerated into a set of full ligands, some of which may become hit candidates if they show a high docking score and adequate location inside the target receptor. The selection of high-scoring MEL fragments with favorable receptor-ligand interactions is a crucial step of V-SYNTHES. Our newly developed CapSelect algorithm is an automated selection tool, identifying potentially promising MEL fragments, i.e., fragments with sufficient space to grow during enumeration. The CapSelect tool, now applied to a much bigger REAL Space library (21 Bln) is a key part of the V-SYNTHES 2.0 version, that demonstrated better enrichment in computational benchmarks and is already being applied to hit discovery for several therapeutic targets.