The problem of Super Structure Search
User 7c5047cd7b
13-03-2006 06:17:49
I pasted my queries in the word file to explain my question.
Thank you for your help
ChemAxon a3d59b832c
13-03-2006 07:05:45
Dear Alex,
I searched both queries on our online database using superstructure search(table nci) and found the target with both. (The found target id was 49138.)
http://www.chemaxon.com/jchem/examples/jsp1_x
Can you send us the source of the first query molecule which did not produce a hit?
Best regards,
Szabolcs
ChemAxon a3d59b832c
13-03-2006 07:40:05
Alex,
Did you use the online installation on chemaxon.com mentioned above or your own installation? If it was your own, did you use any custom standardization on the table?
Szabolcs
User 7c5047cd7b
13-03-2006 08:05:55
Szabolcs,
Thank you for your help.
I tested both of them on your online database too. (NCI)
But I didn't get the target.
I pasted the screen shot.
The first screenshot is my query.
Please pay attention the third screenshot.
(I thought 49138 should be at the first line of the results.)
Best regards
Alex
ChemAxon a3d59b832c
13-03-2006 08:29:01
Dear Alex,
Thanks, now I can see that the hit is indeed missing. (Earlier I used the id field under Conditions to restrict the search to structure 49138. It seems that the other hits are interfering the search process somehow.)
This is definitely a bug. We will check this issue and get back with the progress.
Best regards,
Szabolcs
ChemAxon a3d59b832c
13-03-2006 14:48:20
Dear Alex,
This bug has been fixed. The next JChem version (3.1.6 or 3.2) will work properly. Thank you for sending us this bug report.
Best regards
Szabolcs
User 7c5047cd7b
17-03-2006 01:14:57
Dear Szabolcs
Thank you for your help.
Is it a bug of Super Structure Search or others?
Do you have any workaround?
Best regards
Alex
ChemAxon a3d59b832c
17-03-2006 08:41:12
Dear Alex,
Yes, it was a bug in the structure searching code. It was a re-initialization problem, and it was working correctly when only one target was used. (E.g. when the compound id was typed in the id field.)
A possible workaround may be to switch off stereo searching. (The double bond stereo information was incorrectly re-initialized within searching.)
However, we will soon issue JChem 3.1.6. (Probably next week.)
Best regards,
Szabolcs
User 7c5047cd7b
17-03-2006 10:51:42
Dear Szabolcs,
I pasted some structures. (Query1.mrv, Query2.mrv, Target.mrv)
I could find the Target.mrv using Query1 and Query2 by Sub SS.
But I couldn't find Query1 and Query2 by Super SS.
If it is the same reason, hope it help you test the new verion.
Best regards
Alex
ChemAxon a3d59b832c
17-03-2006 13:50:53
Dear Alex,
This seems to be the same bug. It will be fixed in the next release.
Best regards,
Szabolcs
User 7c5047cd7b
18-04-2006 05:05:13
Dear Szabolcs
Would you help me check the following case?
I pasted the structures. (query.mrv, target.mrv )
(The version of jchem is 3.1.6)
The target.mrv is saved in a database of jchem.
I used the query.mrv to search the database by SuperSS.
I don't think the target.mrv should be hit,
because there are some query properties.
I would like to know the reason.
Best Regards
Alex
ChemAxon a3d59b832c
18-04-2006 09:32:38
Dear Alex,
Please note that query structures are not supported in the database. You may be able to insert these structures and retrieve from the cd_structure column, but not all query features are handled internally. (Some even throw exception, I guess.)
In the future we may introduce query tables.
Best regards,
Szabolcs
User 7c5047cd7b
18-04-2006 10:19:25
Dear Szabolcs
Thank you for your reply.
I pasted another mol file, which don't have the query features.
Please test this one.
My question:
Doesn't Jchem database support query structures?
I tested some cases. It worked well.
Best regards
Alex
ChemAxon a3d59b832c
18-04-2006 12:51:15
Dear Alex,
| AlexK wrote: |
I pasted another mol file, which don't have the query features.
Please test this one. |
Here the problem is that default standardization removes explicit hydrogens from the database molecules. If you use a custom standardization with only aromatization action, the database structures with explicit hydrogens will work as you expect during superstructure search. | AlexK wrote: |
My question:
Doesn't Jchem database support query structures?
I tested some cases. It worked well. |
JChem is not prepared for the storage and the handling of query features on the database side during searching. Some features may work accidentally, but don't be surprised to find errors in this area.
Why it looked OK so far: If you could import your queries into the database without error, the query features are reserved in the cd_structure column, as it contains your original input. So these can be retrieved unchanged. However, searching uses a different, standardized column (cd_smiles) that may miss some of the query features.
Best regards,
Szabolcs
User 7c5047cd7b
19-04-2006 02:02:18
Dear Szabolcs
| Quote: |
| Here the problem is that default standardization removes explicit hydrogens from the database molecules. If you use a custom standardization with only aromatization action, the database structures with explicit hydrogens will work as you expect during superstructure search. |
I tested to use a custom standardization with only aromatization.
But I couldn't get the correct result.
Alex
ChemAxon a3d59b832c
19-04-2006 07:26:26
Dear Alex,
I could not reproduce the problem. (For me, Query.mrv does not hit Target1.mrv in the database in superstructure mode.) It is possible that the other structures in the database have some effect on this bug. Is it possible to export and send the structures from your table? (You can send it in email, if sensitive information.)
Best regards,
Szabolcs
User 7c5047cd7b
24-04-2006 03:18:50
Dear Szabolcs
Sorry for my delay this case.
You are right.
The explicit hydrogens worked well in SuperSS,
after I used a new standardize XML.
Thanks
Alex