SubStructure search : MarvinSketch+JChemSearch vs jcsearch

User a68e2d9365

03-10-2007 13:17:18

Hello,





I'm using at the same time 2 tools to perform the same SubStructure queries :





1) I imported the TimTec_STOCK2004V06.sdf catalog file within chemaxon tables in an Oracle Database. I use MarvinSketch 4.1.5 to load the query structure then I use JChemSeach java class to run the SubStructure query against my Oracle Jchem structure table.





2) I run queries directly against the sdf file TimTec_STOCK2004V06.sdf. I


use the jchem3.2.3/bin/jcsearch on line tool.





problem : when I load the following structure (see file attached) as query I get :


- 114 hits using MarvinSketch,


- 160 hits using jcsearch.





File format used for both queries is sdf. (not smiles - when I use smiles I get only 1 hit but it comes from the Hydrogen interpretation I suppose).





I checked that the 160 hits I found were correctly loaded within the Oracle chemaxon structure table. So I'm lacking 46 hits. Any hint to help me ?





jcsearch command line :


/production/chemaxon/jchem3.2.3/bin/jcsearch -t:s --or -q /production/HitDBProd//tools/sssearch/in/3pyro.sdf -f sdf -o /production/HitDBProd//tools/sssearch/out/20071003_AML_test24/TimTec_STOCK2004V06-hits.sdf /production/chemaxon/comercial_compound/TimTec_STOCK2004V06.sdf





Alain MEIL.

ChemAxon a3d59b832c

03-10-2007 20:06:49

Hi Alain,





Exactly how did you specify the query for JChemSearch? As SDF source String or as Molecule object? Or maybe converted to SMARTS?





Best regards,


Szabolcs

User a68e2d9365

04-10-2007 09:22:58

Hello,





My mistake,





After checkings, I finally found the 160 hits from MarvinSketch+JchemSearch.





After performing the search I got 1291 hits on all my catalogs then I had to make a join to filter the Timtech ones. The first 1000 rows returned 114 hits within TimTech catalog. (I use Oracle 'IN' clause which allows only 1000 entries). Within the 291 rows remaining I got 46 hits within TimTech (114+46=160).





Sorry to have bothered you.





Regards,





Alain.

ChemAxon a3d59b832c

04-10-2007 09:37:00

We are relieved, thanks. :)





Szabolcs