O.K. More info now. It seems like MolFromSmiles works well. The problem is
with MolToSmiles. Perhaps the ring closure marker digit is causing the
problem. Attached is a simpler example. The input smiles is
'O/C=C1\\CCCC\\C1=N/F'.
The bond info of the created molecule looks healthy. But MolToSmiles gives
a different molecule.
**** By the way, how can I reply to my own thread without creating a new
thread each time and thus messing up this forum? **** I don't receive an
incoming email for posts that I made and hence I cannot reply to an
incoming email.
Thanks.
Ling
Ling Chan <lingtrek...@gmail.com> 於 2021年11月6日週六 下午1:05寫道：
> Dear colleagues,
>
> I have encountered an instance where MolToSmiles(MolFromSmiles()) resulted
> in a change of stereochemistry of a double bond. An attached figure
> illustrates this.
>
> The problematic input smiles is
> "C1Nc2ccc(C)cc2/C1=C1/Nc2ccccc2/C1=N\OCCOF".
>
> The original smiles was even longer. I am afraid that this is already the
> shortest version I manage to get. When I cut the methyl group
> ("C1Nc2ccccc2/C1=C1/Nc2ccccc2/C1=N\OCCOF"), or just shorten the long tail
> ("C1Nc2ccc(C)cc2/C1=C1/Nc2ccccc2/C1=N\OC"), things would go well, as shown
> in the other two figures.
>
> It does not seem to be related to the reusing of the ring closure digit
> markers. The problem persisted when I avoid reusing the digits.
> "C1Nc2ccc(C)cc2/C1=C3/Nc4ccccc4/C3=N\OCCOF"
>
> Thank you for your attention. Hope you have had a good weekend by the time
> you see this.
>
> Ling
>
>
>
>
_______________________________________________
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
