InChI 1.07 polymer problem 20260309

[andreierin]

Hello,
Looks like version 1.07 still has problems with polymer encoding and conversion from InChI to structure.
See attached DOCX or PDF.

Best wishes,
Andrey Yerin

Poly 106 vs 107.docx

Poly 106 vs 107.pdf

[flange-ipb]

Dear Andrey,

I'd like to make you aware that the chemical structure editor which is used in the InChI Web Demo - Ketcher - has an issue in serializing Molfiles for polymers. I reported those problems a few years ago in the Ketcher repository (epam/ketcher#2999) and the underlying cheminformatics framework Indigo (epam/Indigo#2174) but nothing has been fixed so far.

I can still reproduce the examples mentioned in those tickets at the InChI Web Demo. I'm also able to reproduce your first drawing and the resulting wrong InChI, so Ketcher/Indigo might be the root cause of your issue and not InChI itself.

If you want to inspect the Molfile from Ketcher which is sent to the InChI backend you can execute await getKetcher("inchi-tab1-ketcher").getMolfile() on the browser's JavaScript console (inchi-tab1-ketcher is the Ketcher frame in the "Draw structure and convert to InChI" tab). This Molfile is different from the one you see when hitting Ketcher's "Save as ..." button. Sadly, I never managed to extract the latter "good" Molfile from Ketcher in an automatic way.

So in conclusion, I would suggest to always crosscheck InChI generation with Molfiles under your control in the "Convert Molfile to InChI" tab of the InChI Web Demo. Regarding your example, you can also take the "good" Molfile from Ketcher and paste it there to receive the InChI 1.06 result.

Best regards,
Frank

[fbaensch-beilstein]

@andreierin As Frank already said at least the difference in the InChI for the first polymer is a Ketcher based problem. The SBL field is missing in the "internal" mol file.
To get rid of any Ketcher smells the best way is to use the "Convert Molfile to InChI" tab.

@flange-ipb Thanks!