Need structural key fingerprint from a query molecule

Greetings ChemAxon Forum Users,

     I work in support of Dr. Richard Judson of the U.S. EPA which produces the ACToR web page (online at actor.epa.gov).  We are trying to use ChemAxon software to compare a query molecule to molecules in our database based on a fingerprint of structural key bits (one bit per key).  I have succeeded in using JChemManager to build a set of fingerprints with both path-based bits and structural key based bits for our database of chemicals.  Now I just need to make a tool which will ultimately be called by a web-based query, get the fingerprints, and use some MySQL queries to compare the query molecule to our database of chemicals.  It appears that GenerateMD should produce the fingerprint, but I do not know the right parameters to use to specify the Structural Key list.  I see in the following parameters in the JChemProperties table in MySQL:

actor_keys.fingerprint.numberOfBits         512
actor_keys.fingerprint.numberOfOnes           2
actor_keys.fingerprint.numberOfEdges          6
actor_keys.fingerprint.numberOfStrucFPCols    1
actor_keys.fingerprint.structuralKeys         smartsMode=false<#*%*#>c1cccc1<#%#>C(=O)O...

But it is not clear how these map to configuration information in cfp.xml such as that on the following line:

<Parameters Length="512" BondCount="8" BitCount="2"/>

If someone could point me in the right direction of the correct parameter names and/or offer a way to specify the structural keys in a java program using the GenerateMD class, we would be most grateful.

thanks very much for any suggestions,
Tom Transue
--
Thomas R. Transue, Ph.D., Computational Biochemist
High Performance Computing & Scientific Visualization
Lockheed Martin Information Technology, Supporting the EPA
Research Triangle Park, NC
919-541-0494

Federal Infrastructure Contact - Ravi Nair
919-541-5467      - nair.ravi _at_ epa.gov
Federal Visualization Contact - Joe Retzer, Ph.D.
919-541-4190      - retzer.joseph _at_ epa.gov

Dear Thomas,

GenerateMD unfortunately cannot generate structural keys, the 'Structural Keys' is a JChem Base feature only at the moment. By the way, the molecular descriptor configuration file does not contain map for structural keys.
The Structural Keys which generated by JChem Manager are usable for the default fingerprints (Chemical Hashed Fingerprints) in JChemSearch. Pleas see the documentation about it: http://www.chemaxon.com/jchem/doc/dev/dbconcepts/index.html#structuralKeys. />However if you don't want to use molecular descriptor fingerptrints for screening, JChem Base provides you the Chemical Hashed Fingerprints for each molecules, the structural keys, and the search functionalities.

Let me know if any other question appeared,

Best Regards, Edvard

Hi Thomas,

 

If not confidential could you let us know how exactly are you planning to do the fingerprint comparison and what is your goal ? Will you calculate similarity or it has some other purpose ?

Edvard is right that the built-in chemical fingeprint  (CFP) implementation in generateMD does not generate the structural key part of the fingerprint, only the CFP part.

You could implement this as a custom descriptor for generateMD, but it might be an overkill.

Alternatively you c an also generate the CFP part and add the structural key part. You can simply use MolSearch to test if a key is a substructure of a query or not and set the corresponding bit to 1. Make sure you standardize the molecules with the same Standardizer configuration that is set for you DB table and that you process the same keys in the same order.

If we can find out more about your aim and this solution is relevant for you we can provide some more information.

 

Best regards,

Szilard

 

 

 

 I have succeeded in using JChemManager to build a

 set of fingerprints with both path-based bits and structural key based 

bits for our database of chemicals.  Now I just need to make a tool 

which will ultimately be called by a web-based query, get the 

fingerprints, and use some MySQL queries to compare the query molecule 

to our database of chemicals.

Hi Tom,

 

I recommend to have a look at one of our web-based application examples.

For example:

http://www.chemaxon.com/jchem/examples/db_search/index.jsp

The source of this application is part of the JChem distribution. See:

http://www.chemaxon.com/jchem/examples/

 

I hope this helps.

Best regards,

Szabolcs