Combining traditional 2D and modern physicalorganic-derived descriptors to predict enhancedenantioselectivity for the key aza-Michaelconjugate addition in the synthesis of Prevymis?(letermovir)

摘要

Quantitative structure–activity relationships have an extensive history for optimizing drug candidates, yetthey have only recently been applied in reaction development. In this report, the predictive power ofmultivariate parameterization has been explored toward the optimization of a catalyst promoting an azaMichael conjugate addition for the asymmetric synthesis of letermovir. A hybrid approach combining 2DQSAR and modern 3D physical organic parameters performed better than either approach in isolation.Using these predictive models, a series of new catalysts were identified, which catalyzed the reaction toprovide the desired product in improved enantioselectivity relative to the parent catalyst.