MolExporter doesn't return standard InChI by default?

User 57295192cc

11-04-2013 13:26:25

Hi,


After upgrading from 5.12 to 5.12.2 we found a difference in behaviour: when exporting to InChI, 5.12 always returned strings starting with "1S/". 5.12.2 returns strings starting with "1/".


I found a forum topic which may be related, but I'm not entirely sure: https://www.chemaxon.com/forum/viewpost47614.html&highlight=inchi#47614


I also found that I can get standard InChI by adding the "SAbs" flag. Is this change intentional, and should we add "SAbs" if we expect standard InChI?


Many thanks,


Pál

ChemAxon 712bc8fcf4

12-04-2013 07:12:42

Hello Pál,


Can you attach the molecule that you want to export?


Regards,
Máté 

User 57295192cc

12-04-2013 09:11:56

Hi Máté,


Here is a little test case:



import chemaxon.formats.MolExporter;
import chemaxon.license.LicenseManager;
import chemaxon.naming.document.TextToStructure;
import chemaxon.struc.Molecule;

public class InChITest
{
    
    public static void main(String[] args) throws Exception
    {
        LicenseManager.setLicenseFile("/tmp/jchem_license.cxl");
        Molecule molecule = TextToStructure.extract("Methanol", "-cas,-elements,-ions,+vernacular,timeout=30").get(0);
        System.out.println(MolExporter.exportToFormat(molecule, "inchi:AuxNone,Woff"));       // -> InChI=1/CH4O/c1-2/h2H,1H3
        System.out.println(MolExporter.exportToFormat(molecule, "inchi:AuxNone,Woff,SAbs"));  // -> InChI=1S/CH4O/c1-2/h2H,1H3
    }

}



Many thanks,


Pál

ChemAxon 25dcd765a3

15-04-2013 10:58:14

Hi Pál,


You are right we have changed the InChi export to generate standard InChi if the molecule has absolute stereo flag set.


However, we do not set this flag if the molecule has no stereocenter at all, which resulted in a non standard InChi. We will fix this in 6.1 release.


Thank you for the report. 

User 3d3f3767a2

30-05-2013 16:36:36

Having a bit of an issue with this one, as I don't think you can go back a version of a JChem database can you? Is it possible to get a beta or early release with this fix?


Thanks!

ChemAxon fc046975bc

31-05-2013 11:05:30

Hi,


We can include this fix in version 6.0.1. It will be available in a month.


Regards,


Peter

User 3d3f3767a2

24-07-2013 16:01:45

Hi there,


This problem seems to still exist in 6.0.3. For example if you use this SMILES:


CC(=O)OCC1OC(OC2=CC=C(O)C=C2)C(O)C(O)C1O


In returns a non-standard InChI, even on your own web services REST demo (http://restdemo.chemaxon.com/#stringMolExport).


Any ideas?



Thanks,
Craig 

ChemAxon d26931946c

24-07-2013 16:24:42

Hi,


 


The SMILES you've sent contains two undefined stereocenters.


I've marked them in this: CC(=O)OC[C@H]1O[C@@H](OC2=CC=C(O)C=C2)C(O)C(O)C1O


We generate standard InChi only for molecules with the CHIRAL flag set or without any stereocenters. If the molecule contains undefined stereocenters we don't export it as standard InChi. I think it's ok this way, please explain us why it's not if you think otherwise.


 


BRs,


Peter

User 3d3f3767a2

24-07-2013 17:02:33

Ah yes I see. Sorry, stupid question!