Known drug space as a metric in exploring the boundaries of drug-like chemical space.

In this work, marketed drug compounds (or known drug space) were used as a metric to test the principles of eliminating parent structures of the nitrenium ion (aryl-amine/nitro compounds) as well as sulphur and halogen containing molecules from screening compound collections. Molecules containing such moieties and/or atoms have biological and physiochemical properties, which possibly make them less attractive as leads in drug development. It was found that precursors to the nitrenium ion were relatively abundant in known drug space at 14%. Thus, their simple elimination from drug-like chemical space is not advisable. Interestingly, the mutagenic potential of the nitrenium ions is linked to their stability and quantum mechanical calculations can be used to estimate it. Furthermore, 24% of drugs investigated contained sulphur atoms and around 28% were halogenated. As some sulphur containing moieties were abundant whilst others were scarce, it was deduced that it would be more effective to eliminate specific molecular scaffolds rather than all sulphur containing molecules. In conclusion, it has been shown that by statistically analysing known drug space a better understanding of the boundaries of drug-like chemical space was established which can help medicinal chemists in finding rewarding regions of chemical space.