SMILES conversion

  Hi,

 During the process of loading compounds into our database from sdf files we convert the standardized molecules to smiles format. We have noticed a problem occurs before we standardize though when the molecules have stereocenters where the stereo bonds have been drawn incorrectly. This issue can be flagged in Marvin and the Structure Checker correctly recognizes that the wedge bonds are invalid and corrects them properly. However when we use molecule.toFormat("cxsmiles:-leu"), the invalid stereo bonds are removed and we obtain a smiles with no stereo centers and importing into Marvin gives us a flat molecule. If we use molecule.toFormat("mol") and import the result into Marvin, the invalid stereo centers are preserved and can be corrected. I have attached an sdf file with some example compounds where we have this issue.

There is no warning given that stereo chemical information is being lost during the molecule to smiles conversion. Is there are way to get notified that this is occurring so that we could trap the errors rather than loading the incorrect smiles strings and then having to track these molecules down for correction? 

final MolImporter molImporter = new MolImporter(new FileInputStream(getSdFile()));

 Molecule molecule;

while ((molecule = molImporter.read()) != null) {

 LOGGER.debug("After reading " + molecule.toFormat("mol")); //stereochem kept

LOGGER.debug("After reading " + molecule.toFormat(ChemUtils.SMILES_FORMAT)); //stereochem lost

Thanks

Cheers

Bob

 

Hi,

There

is no warning given that stereo chemical information is being lost

during the molecule to smiles conversion. Is there are way to get

notified that this is occurring so that we could trap the errors rather

than loading the incorrect smiles strings and then having to track

these molecules down for correction? 

Actually there are no stereo chemical information at the attached molecules as the wedges are drawn to such atoms which cannot be stereo center. So it is not possible to drop warning message in this case.

Andras

Bob,

How about using structure checkers API?

That could notify you about the errors, and even fix them, too.

 

BR,

Szabolcs

Hi Szabolcs,

Yes, I think that would be the best option for us as we could fix several of the other annoying errors we encounter as well with the Structure Checker API.

Thanks for your help

Cheers

Bob