Generating SMILES without Chirality Information

[Jesse-zjx]

I'd like to express my gratitude for your outstanding work on this project. It has been incredibly helpful in my research.

However, I noticed that the generated SMILES strings do not include chirality information, which makes them somewhat inconsistent with uspto-50k test set.

Could you please provide some guidance on how to generate SMILES strings with the appropriate chirality information? I would like to eval the results with Chem.MolToSmiles(mol, isomericSmiles=True).

Again, thank for your outstanding work~

[arneschneuing]

Hi @Jesse-zjx, thanks for your message and sorry for the slow response. Unfortunately, I'm afraid this is not possible because we never used chirality information in this project.

[mrwnmsr]

also, note that uspto50k has a considerable proportion of stereochemistry errors, which e.g. prohibit models to ever reach 100% accuracy on the dataset

[Jesse-zjx]

@arneschneuing @mrwnmsr Thank you for your patience in answering. BTW, could you provide references or sources for "uspto50k has a considerable proportion of stereochemistry errors"?

[mrwnmsr]

source: own analysis ;)

[Jesse-zjx]

LOL, thank you for your analysis of the stereochemistry in 50k, which provides some basis for our following research.

Is it possible to provide your methods/ideas or tools for analyzing stereochemistry? This will be of great help to us.