The Chemical Catalysis program in the Division of Chemistry is funding the project "Computational Prediction of Enantioselectivity in Metal-Catalyzed Reactions" in the groups of Olaf Wiest and Paul Helquist at the University of Notre Dame. The project develops, validates, and applies computational methods for the prediction of stereoselectivity (preferential formation of one arrangement of atoms in space over other possibilities in a molecule) and regioselectivity (preferential formation of new bonds at specific positions of a molecule over other possible positions). The ability to correctly predict reaction selectivity in transition metal catalysis has a broad impact in the pharmaceutical and fine chemicals industry. Highly selective reactions are needed to produce safer, more effective, and lower cost drugs. The project provides multi-disciplinary training for graduate and undergraduate students that emphasizes the close connection between computational and experimental approaches and their application to practical problems.
The prediction method involves five steps: (i) calculation of the pertinent transition structures for the targeted reactions using electronic structure methods, (ii) parameterization and validation of ground- and transition state force fields (TSFF) using quantum-guided molecular mechanics (Q2MM), (iii) calculation of the free energy difference of the relevant diastereomeric transition structures using full Monte Carlo conformational sampling at the transition state with the TSFF, (iv) use of the force fields for rapid prediction of enantioselectivities for virtual ligand libraries, and (v) selection and experimental evaluation of top ligand candidates for validation. The methods are used for a variety of reactions of importance in synthetic organic chemistry, including: (i) Mukaiyama aldol condensations, (ii) Heck arylations, (iii) 1,4-addition of organoboron compounds to unsaturated carbonyls, and (iv) imine reductions. The TSFFs and the Q2MM code is available to the scientific community via Github.
