Presentation description

A Ni catalyzed diastereoselective synthesis of tetrasubstituted alkenes was previously reported, and data science tools can be applied to further understand this reaction. In particular, multivariate linear regression (MLR) has been proven a useful tool in reaction development as it can reveal correlations between reaction outcomes and ligand properties, which can provide mechanistic insight as well as aid in reaction optimization. This work seeks to build MLR models for the previously reported Ni-catalyzed cross-coupling reaction based on catalytically relevant transition state and intermediate structures, with the goal of parametrizing models that are as informative and accurate as possible. To do so, the necessary structures were first generated, optimized, and filtered. Energies and properties of interest were then calculated for these structures, and the data was subjected to a nested cross-validation scheme to identify the best models. This process was applied to ligands, an intermediate in the catalytic cycle, and the diastereoselective transition state. MLR models have been parameterized and further work is needed to better interpret and apply these models.