tautomer canonicalization

Is there currently a way to generate one (the most stable) canonical (standardized) tautomeric form for an input molecule, independent from the tautomeric form of the input structure, i.e. canonicalize tautomers?





Thanks,





Stephan

The most stable tautomer calculation is just under development, so that canonical tautomer form generation will be available soon.

Thanks. This will be very valuable! -S.

Tried to find the canonical tautomer of abacavir - so that I can collapse all of my abacavir records into one compound. I am using the latest 3.2 release. I notice that several forms of this same molecule do not canonicalize to one single structure. I have found a work-around for this particular example, but worry that it does not generalize.





Initially, I just want to use cxcalc at the command line:





C:\>cxcalc "NC1=NC(NC2CC2)=C3N=CN([C@@H]4C[C@H](CO)C=C4)C3=N1" canonicaltautomer


id structure


1 OC[C@H]1C[C@H](C=C1)n2cnc3c(NC4CC4)[nH]c(=N)nc23





C:\>cxcalc "OC[C@H]1C[C@H](C=C1)n2cnc3c(NC4CC4)[nH]c(=N)nc23" canonicaltautomer


id structure


1 Nc1nc2n(cnc2c(=NC3CC3)[nH]1)[C@@H]4C[C@H](CO)C=C4





C:\>cxcalc "Nc1nc2n(cnc2c(=NC3CC3)[nH]1)[C@@H]4C[C@H](CO)C=C4" canonicaltautomer


id structure


1 OC[C@H]1C[C@H](C=C1)n2cnc3c2[nH]c(=N)[nH]c3=NC4CC4





You'll notice in particular that #2 and #3 canonicalize to each other, and not to a common structure.





So instead, I tried using the tautomerPlugin directly, and then find the lowest string amongst all the tautomer forms for each molecule. This seems to give me one unique structure from all of the tautomer forms above. I didn't set any of the parameters for the plugin. I thought setting 'canonical' or 'dominant' would help, but I get the same problem above. Instead, I went the other route of enumerating all tautomer forms and comparing strings.





if (tautomerPlugin == null) {


tautomerPlugin = new TautomerizationPlugin();





// set plugin params


Properties params = new Properties();


//params.setProperty("single", "true");


//params.setProperty("pH", "7.4");


//params.setProperty("type", "structure");


//params.setProperty("max", "1");


//params.setProperty("canonical", "true");


//params.setProperty("dominants", "true");


tautomerPlugin.setParameters(params);


}





tautomerPlugin.setMolecule(mol);





tautomerPlugin.run();


if (tautomerPlugin.getStructureCount() > 0) {


Molecule taut = tautomerPlugin.getStructure(0);


taut.aromatize();


String canonical = taut.toFormat("smiles:u");


int best = 0;


for (int i=0; i<tautomerPlugin.getStructureCount(); i++) {


taut = tautomerPlugin.getStructure(i);


taut.aromatize();


if (taut.toFormat("smiles:u").compareTo(canonical) < 0) {


canonical = taut.toFormat("smiles:u");


best = i;


}


}


return tautomerPlugin.getStructure(best);


}


return mol;

Hi,





I see you use JChem 3.2


New version (3.2.1) of JChem can be excepted soon.








Marvin 4.1.3 release (this will be included with JChem 3.2.1) is available in the test section:


http://www.chemaxon.com/test/marvin


This new test version was released two days ago.








At this site you can try the canonicalization of abacavir. (I tried, and it works well.)











Your method to find canonical form seems to be interesting! I will check more thoroughly and inform you from the results.





Thanks.


Jozsi

Thanks for the update. I'll wait for the newest release.