A Platform for Arene Activation via Machine Learning and Small Molecule Catalysis

/ABSTRACT The development of pharmaceutical compounds and their late-stage analogues is crucial for the advancement of human health through the synthesis of novel pharmaceuticals. Arene rings are especially prevalent in pharmaceutical compounds, making their derivatization highly desirable. Arene epoxidation, once considered a deleterious side product of metabolism, has the potential to be reimagined as a useful mechanism of arene activation under this proposed training plan. Notably, small molecule catalysts able to achieve this transformation are lacking within the literature, necessitating the development of a novel catalyst. Vast data sets within the literature for alkene epoxidation can be utilized as a starting point for catalyst design, but this abundance of data can be challenging to manipulate alone. Furthermore, machine learning has emerged as a valuable tool for the analysis of large data sets, and increasing interest has been applied towards its use for catalyst design and development. Utilization of transfer learning models would therefore enable known literature data to be applied towards a novel catalytic system. Achievement of this catalyst design through machine learning would unlock a platform for activation of a diverse array of arene building blocks for synthetic and medicinal chemistry. Therefore, the specific aims of this proposal are in alignment with the goals of the NIH in the advancement of human health and society. The first Aim of this proposal is to design an arene epoxidation catalyst using transfer learning from native enzyme reactivity. Aim 2 endeavors to optimize the initial catalyst through transfer learning from olefin epoxidation, and Aim 3 will utilize a recommender model and transfer learning to achieve the derivatization of the arene oxide intermediate into value-added products. This research proposal aligns with the fellowship goals by requiring new skills in the area of computational chemistry, machine learning, and catalyst design that complement my previous training in synthetic organic methodology. The Elkin lab provides the ideal research environment to complete this training, as their research focus lies in the area of machine learning for the development of predictive models for transition metal catalysis and total synthesis. Furthermore, additional guidance from Professor Steve Buchwald will be provided, and he has expressed his support and dedication to the scientific and professional development goals of this training plan. The Buchwald lab’s extensive expertise in ligand design and catalyst development will be instrumental in the execution of this training plan, especially in my catalyst optimization campaign. With the goal of securing an academic position at a research university in the future, the Elkin group with the additional support of Professor Buchwald will represent the ideal training grounds for this career goal. Furthermore, MIT is one of the most productive research institutions with ample resources and equipment in order to achieve the proposed research training plan. Project Number: 1F32GM163368-01 | Fiscal Year: 2026 | NIH Institute/Center: National Institute of General Medical Sciences (NIGMS) | Principal Investigator: Andria Pace | Institution: MASSACHUSETTS INSTITUTE OF TECHNOLOGY, CAMBRIDGE, MA | Award Amount: $74,632 | Activity Code: F32 | Study Section: Special Emphasis Panel[ZRG1 F04A-M (20)] View on NIH RePORTER: https://reporter.nih.gov/project-details/11317766