Identification of catalysts

[dswigh]

There is a list of catalysts in Therapeutics Data Commons (TDC) that could potentially be used to identify catalysts (https://tdcommons.ai/multi_pred_tasks/catalyst/), in a similar way to how we identify solvents. However, identifying catalysts by using a list is not necessarily a good idea, since the same set of organic molecules could play many different roles, depending on the context (so, identification of solvents by list is probably less error-prone than identifying catalysts like this). One way to improve catalyst identification would be to segment the catalyst list by reaction class, such that only a subset of all catalysts are associated with a particular reaction class (potentially by clustering catalysts and associating clusters of catalysts with rxn classes from name rxn), however, this matching process would be very laborious and go beyond the scope of ORDerly, which is built to be primarily computationally extensible and not have hand-crafted rules.