Abstract:
In order to improve the success rate of new drugs, major pharmaceutical companies are committed to developing new strategies and technologies for Computer Aided Drug Design (CADD). Scaffold hopping is one of the most commonly used strategies for discovering lead compounds. However, most of the lead compounds generated by existing scaffold hopping strategies are derivatives or analogs of reported lead compounds that lack novelty. To improve structure novelty, we proposed a geometry-based scaffold hopping strategy, which took the idea for the first time to keep the relative distances and angles unchanged between pharmacophores and skeleton. Thus, a linker library for scaffold hopping based on geometric features was constructed. The original compounds of the linker library were derived from Cambridge Structural Database (CSD), and which were broken into fragments by Retrosynthetic Combinatorial Analysis Procedure (RECAP). After a series of screening and processing, we designed a linker library containing 1587020 fragments with 2~4 connection points. There was a case about Skepinone-L, a selective inhibitor of MAPK11, which verified the usefulness of the linker library. Based on the distances and angles between the center to connected points of the skeleton, the linker library could recommend fragments with novel scaffolds to maintain the relative position of pharmacophores, which is beneficial to guide scaffold hopping and accelerate the process of drug discovery.