Cheminformatic Analysis of Core-Atom Transformations in Pharmaceutically Relevant Heteroaromatics
J. Med. Chem., 2025, 68, 6, 6027–6040
The art of medicinal chemistry involves discovering molecules that represent a compromise between target potency and in-vivo pharmacokinetic properties. Only after a successful balancing act between these two (sometimes conflicting) factors has been achieved, is it possible to propose a candidate drug. This is a particular challenge when the required route of administration is through oral dosing.
It’s for this reason that heterocyclic organic molecules are both the language and lifeblood of medicinal chemistry. Being able to try carbon → heteroatom substitutions at precise locations on a molecule facilitates the testing of hypotheses and can lead to breakthroughs in drug discovery.
A common strategy in medicinal chemistry campaigns is to search for “isosteric” (same size or shape) groups, regarding a specific part of the problem molecule in order to find one with optimal properties. An ideal way to make analogues of problem molecules is by late-stage functionalisation (LSF) of the molecule itself, avoiding de novo synthesis each time and saving valuable lab time and resources. In recent years, skeletal editing of heterocycles has emerged as an intriguing approach to the interconversion of heterocycles as a form of LSF.
In this paper, the authors performed virtual skeletal editing (including known and also as yet unknown synthetic transformations) on open source and proprietary compound libraries. They sought to identify the transformations that would lead to the largest expansion of accessible chemical space, indicating transformations that would have a high impact on compound diversification campaigns.
They found that, generally, ring contraction transformations gave access to more novel / uncharacterised chemical space than ring expansion or atom transmutation transformations. Additionally, multiple atom transformations, e.g. pyrazole → furan, gave more uncharacterised products than single atom transformations. The generated data highlight the transformations that should be developed into practical synthetic methodology if possible.
While expanding accessible chemical space around a problem molecule is a worthwhile goal, an additional suggestion would be to interrogate these known and unknown transformations on the expected changes in relevant medicinal chemistry properties, such as on- and off-target potency or physicochemical properties. This would offer a complimentary outlook on the impact of skeletal editing reactions.