Conversion of the Chemistry by Design Database into a Machine-Readable Format for Training and Testing Machine Learning and Artificial Intelligence Models

Abstract

This dissertation is organized into six chapters, encompassing both the expansion of our group's native methodologies around indole synthesis and our anionic amino-Cope platform, as well as the exploration and enhancement of our educational outreach initiatives such as the Top 200 Drug Posters and Chemistry by Design. Chapter 1 introduces my graduate research with a new methodology employing a dearomatization-rearomatization approach via Wessely oxidation and LTA. Chapter 2 shifts focus to a brief biological collaboration that entailed the total synthesis of (-)-iridomyrmecin. The objective was efficient production rather than developing a novel synthetic route, thus existing synthesis strategies were followed, leading to some inventive methods for diastereomer separation due to limitations in available techniques. Chapter 3 presents a novel entry into the anionic amino-Cope platform, driven entirely by computational results; a first for the Njardarson lab. Chapter 4 provides an in-depth discussion on Chemistry by Design, including its conversion into a machine-readable format for training and testing computer models, detailing the dataset extraction processes, encountered challenges, and solutions. Chapter 5 continues from Chapter 4 and focuses on analyzing the content of the dataset, highlighting significant insights gleaned from accessible data. Chapter 6 concludes with an overview of the current state of the Njardarson Lab’s Top 200 Drug Posters, showcasing the transformation of the creation process into an automated procedure, significantly reducing the production time from days or hours to minutes.