Any, Hetero, Atom List searches not working properly

Hi,





any, hetero, atom list searches are not working properly.





JCHEM BASE 3.1.5, MARVIN 4.0.4, JVM 1.4.1_02





All the information is in the attached document.





Thanks.

We will check these issues.





Do you use the custom standardization configuration that Ann presented on our user meeting?





Best regards,


Szabolcs

Not for this searches.

Hi,

Quote:

Questions about about Plugin Why do the following functions not seem to work in the Marvin Plugin? Is a problem of the plugin or Jchem?

I couldn't quite understand what the "Plugin" terms refers to in this case.





Could you describe in exactly what environment these searches fail ?


E.g. do you use command-line or JChem API ?


If command-line, do you search in a file or in a table ?


If API, what class do you use ? MolSearch or JChemSearch ?


What options do you set for the search ?





We couldn't reproduce any of the problems so far.


For the first three (any, hetero, atom list) one can achieve the same behavior by turning on "exactQueryAtomMatching". Isn't it tuned on in your application ?





Also what file format do you use to pass around structures ?


Could you upload the exact file or string that is passed directly to our API (or application) ?





I know we've already asked more than you :) , but this information is necessary to solve the problem.





Best regards,





Szilard





PS: We generally prefer molecule sources as structure file attachments: it's not just that we don't have to draw them again, but they are also more accurate (e.g. same file format, same atom order, etc.).

I could reproduce the atom list issue. This is due to aromaticity. Please note that in your query with the atom list, the middle ring's aromaticity is ambiguous. In case of C, it is aliphatic (C is sp3) and in case of S it is aromatic. For this reason, the standardization on the query cannot aromatize this middle ring. Therefore, it does not match to the aromatic versions.





One solution could be to formulate your query with "single or aromatic" bonds in the middle ring (see attached), or from the next major version you can use vague bond option, level 2 or above to get all the desired hits.





(See UGM presentation for explanation here: http://www.chemaxon.com/forum/viewpost6491.html#6491 )





For the rest of the problems, I recommend to check the exact query atom matching as suggested by Szilard. If that was set false(default setting), we will need the actual query structure and the molecules from the cd_structure column which were screened by the query. (From chasing bugs earlier I can say that sometimes the order of molecules and order of atoms may be important to reproduce a problem. For this reason, we will need this actual data.) Let us know if you need help getting that information out of the database.





Best regards,


Szabolcs

Hi,





Please disregard the first part of my previous post. I realized that I made a mistake while drawing the molecule you reported. (Sketching based on the pictures in the word file.)





The middle ring in all your examples are always aliphatic, and it should be found. (And it seems to work.)





Sorry for the confusion,


Szabolcs

So we still could not reproduce the problems. Could you send us the screened molecules to further investigate the problem?





Szabolcs

Just for the record:





Follow-up in email revealed that setExactQueryAtomMatching(true) was used for the searches. Removing that should fix searching behaviour.





setExactQueryAtomMatching(true) makes the atom list only match to atom list, "any atom" match to the "any atom", etc. This setting is only useful for duplicate checking of queries.

Hi,





The solution was setting setExactQueryAtomMatching(false), or leaving it to its default value, which is false.





Thank you very much !!!