comparaison substructure search by JCHEM and IDBS tools

Do you know if a comparaison between substructure search by JCHEM and IDBS tools has already be realized?





Now, my application is using IDBS tools and I would like to compare it with Jchem for substructure search.


I install the server tomcat and I use the example given in Jchem directory; then i insert in a Mysql database my 35745 structures and i compare the results given by the 2 tools.


the first experiment gives approximatively the same results.


How run a substructure search by Jchem (by fingerprint? by graph?....)

Hi TRAN-MINH,





We have never compared JChem and IDBS yet. Some of our users have done it as I know, but we haven't got any results about it. The search time appears in our JSP example on upper right corner of result page (Search took x.yy seconds.), it can help you to compare the performance.





A two stage method provides optimal performance during database search in JChem:


- Rapid pre-screening reduces the number of possible hit candidates:


Chemical Hashed Fingerprints are used for substructure and superstructure searches, hash code is used for duplicate filtering (usually during compound registration)


-Graph search algorithm is used to determine the final hit list


You can find information about our structure search technology here: http://www.chemaxon.com/conf/JChem_Base.ppt





Regards, Edvard

I have a database of ~ 35000 structures.


I would like to compare the substructure research by IDBS and JCHEM tools.


I have a problem with a research (see attached file) :


the option used in the 2 case is absolute chemistry.


With IDBS tools, I have no results whereas with JCHEM tools I have 85 results.


Why?


The 85 results from Jchem seems false.


The bond with Cl is single in the query whereas in the results the bond with Cl is up?








If I change my query in putting the up bond with Cl, I have the same results with JCHEm 85 but 7 with IDBS.

I choose in the query the option exact stereochemistry, and I have only 9 results: The structures like the third example in the attached file disappear

Hi,





Can you post the query and an example database structure which is found by JChem as extra?





Thanks,


Szabolcs

Do you not received the file?


I joint it again

with option stereochemistry = on => 85 results


with option stereochemistry = exact => 9 results (the structures like the third example disappear)





What is the difference between the 2 options (I didn't find it in the help)

Hi Tran-Minh,





Thank you for the examples.

TRAN-MINH wrote:

With IDBS tools, I have no results whereas with JCHEM tools I have 85 results. Why? The 85 results from Jchem seems false. The bond with Cl is single in the query whereas in the results the bond with Cl is up?

The stereo configuration of a stereo center can be represented in many different ways. For example, I attach a file to show different wedge/hatch bond representations of your substructure. (stereo_configuration_s.png)





Accordingly, stereochemistry matching in JChem does not directly consider the bond types (wedge or hatch), but rather checks whether the configuration is the same. It seems to me that all the attached results are correct in this manner.

TRAN-MINH wrote:

If I change my query in putting the up bond with Cl, I have the same results with JCHEm 85 but 7 with IDBS.

In agreement with the explanation above, this second query has the same configuration as the previous one, so it is natural that the same results are obtained.





In summary, it rather seems to me that the IDBS search is buggy.

TRAN-MINH wrote:

with option stereochemistry = on => 85 results with option stereochemistry = exact => 9 results (the structures like the third example disappear) What is the difference between the 2 options (I didn't find it in the help)

The exact stereo option means that all stereo information should be the same in the query and target ("all stereo info is exactly the same"). It mainly has an effect when the query has no stereo information: it only matches non-stereo target. Similarly, a query with a wiggly tetrahedral center will only match wiggly tetrahedral center, and not specific R and S configurations.





In the case of your query and result 3, the atom denoted by map 1 in the attached picture below (query.png) makes the difference. In the query it has no specified stereo configuration, but its corresponding atom in result 3 has. Therefore the stereo information is not "exactly the same".

TRAN-MINH wrote:

(I didn't find it in the help)

Sorry about that, we will complete the documentation with this explanation.





Best regards,


Szabolcs