Chemical Space Exploration

Klingler et al. introduced the concept of "SAR by Space" (SAR = structure-activity relationship; you may have heard of "SAR by NMR" by Abbott, or "SAR by Catalog" by Astex). This method aims to accelerate early stages of drug discovery by utilizing the accessibility of compounds from vendors. The premise is simple: Starting with a query molecule, scientists can search in vast make-on-demand Chemical Spaces for related compounds and analogs. Molecules of interest can be ordered and synthesized within weeks.
This approach has several perks: Companies without in-house facilities can cost-efficiently advance their projects in a short period of time. Furthermore, the increasing number of commercial Chemical Spaces increases diversity and possibilities to find interesting follow-ups.

Peter Ertl extracted bioactive ring systems and their respective targets from a billion of drug-like molecules resulting in the creation of a ring Chemical Space with almost 40,000 molecular scaffolds.
Rings represent privileged motives in medicinal chemistry as they define the shape of the molecule, are responsible for the orientation of attached groups and often contribute to the binding affinity on their own. Yet sometimes ring replacement (also described as scaffold replacement) is required due to promiscuity of the system, unfavored physicochemical properties, toxicity issues, or intellectual property claims.

BioSolveIT software tackles this problem with the scaffold replacement tool ReCore and infiniSee that performs fuzzy pharmacophore search for distant neighbors to a query molecule.

The Reymond group from the University of Bern in Switzerland made their efforts to enumerate small organic molecules leading to the GDB databases. The numbering of the data sets specifies how many heavy atoms were used for the construction of the Chemical Space. Users can use the sets for virtual screening (e.g. with SeeSAR or FlexX) to search for synthetically accessible compounds.
Enumerated data sets require a vast amount of storage to be processed. Hits from combinatorial spaces on the other hand are built up during the search, thus lowering the computational requirements.