Training, validation and test set used on the work: Organic Reaction Mechanism Classification with Machine Learning It contains 5,000,000 samples of kinetic profiles for reaction mechanisms transforming a substrate S into a product P, catalyzed by a catalyst, cat. These were generated by solving the corresponding ordinary differential equations corresponding to 20 typical mechanisms. These mechanisms belong to four distinct categories: 1) the core mechanism (M1), is the simplest Michaelis-Menten-type mechanism; 2) mechanisms with bicatalytic steps (M2-M5), involve either catalyst dimerization (M2 and M3), or a reaction between two different catalytic species (M4 and M5); 3) mechanisms with catalyst activation steps based on the core mechanism, where a precatalyst requires activation unimolecularly (M6), via substrate coordination (M7) or via ligand dissociation (M8); 4) mechanisms with a variety of catalyst deactivation steps from either catalytic intermediate of the core mechanism (M9-M20). These kinetic samples can be used to train, validate and test a machine learning model.
|Date made available||22 Dec 2022|
|Publisher||University of Manchester figshare|
- mechanistic insight
- catalysis chemistry
- reaction kinetics and dynamics