Approximate quantum mechanical calculations, particularly density functional theory, have been successfully used in the previous few decades to help understand the activity of metalloproteins towards various fundamental reactions important both to humans and other forms of life. Examples include Cytochrome P450 which helps metabolize waste products in the body, nitrogenase which takes part in the nitrogen cycle important for food growth, and hemoglobin and myoglobin which regulate oxygen and nitric oxide transport in the body. However, these theoretical methods often have serious difficulty in the treatment of transition metal containing compounds even smaller than these metalloproteins, making interpretation of mechanism in these systems uncertain. This difficulty, in turn makes it difficult to redesign these proteins, create artificial versions, and to design drug targets for them. Phaseless auxiliary-field quantum Monte Carlo (ph-AFQMC) on graphical processing units and correlated sampling offers an accurate and scalable alternative to traditional methods. This application involves the development of a localized orbital formulation of this technique to push it from one only used on small systems to one used reliably on large systems. Then this method will be used both as a benchmark for more approximate methods and used as a correction to cluster models of these metalloproteins. Using this method, questions regarding the mechanism of oxidation in Cytochrome P450, N-N bond cleavage in nitrogenase, and autoxidation in hemoglobin as well as questions regarding the binding of small molecules such as O2, CO, and NO to the heme of hemoglobin and myoglobin will be answered. The work will be undertaken at Columbia University under the mentorship of Prof. Richard Friesner in the chemistry department, an expert in metalloprotein modeling, in collaboration with Prof. David Reichman at Columbia, an expert in Quantum Monte Carlo. The supercomputing facilities at Columbia and at remote facilities such as Oak Ridge National Lab Leading Computing Facility and NSF XSEDE include ample CPU and GPU resources. The fellowship training plan involves publishing in high impact journals and presenting at conferences for both the theoretical chemistry community and the biochemistry community. It also includes the opportunity to mentor graduate students. The career development plans includes attending workshops organized by the Office of Postdoctoral Affairs as well continuing in a leadership role in the Columbia University Postdoctoral Society.
Despite the important role of metalloproteins in human metabolism, human circulatory systems, and the nitrogen cycle important for food production, there is still great difficulty in simulating chemical systems with transition metals. In this work, we propose to develop a localized orbital implementation of phaseless Auxiliary Field Quantum Monte Carlo (ph-AFQMC) to enable high accuracy theoretical mechanistic studies of the mechanisms underlying the activity of these metalloproteins. These studies will enable the design of new therapeutics and artificial devices to improve human health.