the smiles strings with aromatic ring in the dataset are non-standard

[hengzzzhou]

Hello,

Thank you for your work.

I encountered an issue with the dataset provided at this link: https://zenodo.org/records/7928396. Specifically, after running the script python scripts/prepare_data.py --data_path examples/data_ir.pkl --output_path examples/train, the tgt.txt files generated contain aromatic SMILES strings that appear to be non-standard.

When attempting to convert these SMILES strings to SELFIES strings using the selfies library, errors occur. The problem seems to originate from the "[c]" in the SMILES strings, which does not seem to be a standard representation. For example, manually modifying C=C(C)C(=O)Oc1cc[c]cc1 to C=C(C)C(=O)Oc1ccccc1 resolves the issue and allows the conversion to proceed normally.

I suspect there might be an issue with the script used for generating IR spectra, particularly in how it handles aromatic rings.

I appreciate your attention to this matter and look forward to your response.