Leveraging Computational Techniques and Automated Workflows for Efficient Stereoselectivity Predictions in Asymmetric Catalysis

Asymmetric catalysis is at the forefront of organic chemistry techniques for which the computational chemistry community has sought to develop tools that aid in their approaches. Many of these methods have been developed over the years and involve predicting reaction outcomes such as yield and stereoselectivity. These methods involve a balance between time and accuracy, dependent on the application. Approaches that rely on data, such as Machine learning (ML) methods, are often plagued by a lack of available well-curated training data. Owing to these factors, developing tools that expedite accurate predictive methods is instrumental in closing the gap between speed and accuracy and a necessary effort. This thesis describes the steps taken to accomplish such tasks while automating computational processes to create user-friendly methods that computational and synthetic chemists alike can utilize. Workflows were developed to automate the generation of curated virtual structure libraries, the assignment and generation of atropisomers, the 3D modeling of ferrocene, and attempts to automate previous work involving a computational screening of catalysts, including a workflow that attempts to transition to open-source software with greater functionality. The conformational dependence of various descriptors for bisphosphine ligands was also evaluated. Additionally, a side project involving synthetic work to develop therapeutics for a rare genetic disorder is explained.