Overriding Innate Decomposition Temperatures of an Avibactam Prodrug Precursor Using Data Science-Guided Synthesis

Abstract

Statistical analysis is used to correlate the thermal decomposition temperature of diverse leaving groups of an avibactam prodrug precursor. SMILES strings and Mordred calculated parameters were leveraged to provide a time-efficient workflow for model development. The resulting models were deployed to predict a novel analogue with a higher onset temperature, allowing for an overall safer reagent and proof of concept for the workflow. Interpretation of the descriptors featured in the models and subsequent DFT analysis uncovered univariate trends, providing a deeper understanding of the decomposition pathway. Finally, this workflow enabled the development of a predictive model correlating energy output of the precursor analogs for a more comprehensive assessment.