From Data to Chemistry: Revealing Causality and Reaction Coordinates through Interpretable Machine Learning in Supramolecular Transition Metal Catalysis

Supramolecular transition metal catalysts with tailored reaction environments allow for the usage of abundant 3d metals as catalytic centres, leading to more sustainable chemical processes. However, such catalysts are large and flexible systems with intricate interactions, resulting in complex reaction coordinates. To capture their dynamic nature, we developed a broadly applicable, high-throughput workflow, leveraging quantum mechanics/molecular mechanics (QM/MM) molecular dynamics in explicit solvent, to investigate a Cu(I)-calix[8]arene catalysed C-N coupling reaction. The system complexity and high amount of data generated from sampling the reaction require automated analyses. To identify and quantify the reaction coordinate from noisy simulation trajectories, we applied interpretable machine learning techniques (Lasso, Random Forest, Logistic Regression) in a consensus model, alongside dimensionality reduction methods (PCA, LDA, tICA). Leveraging a Granger Causality model, we go beyond the traditional view of a reaction coordinate, by defining it as a sequence of molecular motions that led up to the reaction.