ORDerly
ORDerly copied to clipboard
Chemical reaction data & benchmarks. Extraction and cleaning of data from Open Reaction Database (ORD)
When trust_labelling=False, we are identifying solvents as any molecule that doesn't contribute atoms to the products AND is part of our solvents_set. However, molecules that sometimes act as solvents can...
How do we check we're only extracting single step reactions?
Atom mapping in the USPTO dataset was done using Indigo over 6 years ago (https://figshare.com/articles/dataset/Chemical_reactions_from_US_patents_1976-Sep2016_/5104873), and better tools for atom mapping have since been created, e.g. rxn mapper (https://onlinelibrary.wiley.com/doi/10.1002/minf.202100138). Even...

Currently in our output dataframe we use a mixture of nans and nones, we should converge on one. possibly nan
Currently we expect that all the solvents can be canonicalised. We should force this to be the case.