Standardizer introduces a valence error

[Kohulan]

We’ve noticed that the ChEBI entry CHEBI:141373
appears to have been modified by the Structure pipeline (although it is unclear whether ChEBI utilises this pipeline). In the current record, the sulfate is rendered as OSH([O-])[O-], i.e., sulfur shown with three single bonds and an explicit hydrogen. This state differs from the structure reported in the original paper.

SMILES: O=C1CC(c2ccc(OSH([O-])[O-])cc2)Oc2cccc(O)c21

The Original structure from the paper looks like this:

SMILES: C1=C(C=CC(=C1)OS(=O)(=O)[O-])C2CC(=O)C3=C(C=CC=C3O)O2

Could you clarify why the pipeline introduces this hydrogen?

We notice this happens when we try to standardise a molecule using the structure pipeline.

Example: C=CC=S[SH]/C=C\CSHC[C]=C
Standarized: C=[C]CSH+C/C=C[SH]/[SH]=C/C=C

Here you could see that S with 3 single bonds, hydrogen and a positive charge, this is a valence error.

[nbosc]

Nothing to do with the chembl_structure_pipeline regarding ChEBI.

I've shared this issue with the chebi curation team and it will be fixed. For any future structure issue in chebi, please refer to this repo: https://github.com/ebi-chebi/ChEBI/issues

Can you please share how you managed to standardise the structure in the example?

[Kohulan]

Hi @nbosc ,

Just to be clear, ChEBi does not use the ChEMBL standardisation pipeline? Or do they have a different one for this?

The example molecule was parsed through RDKit to get the molblock, and then it was standardised using the ChEMBL standardisation pipeline.

Usually, the structure pipeline does throw an error [6], but when you try to standardise it ends up doing that.

we later fixed it manually: C=CC=[S]SC=CCS+[O-]