Duplicate check in Jchem.

I have a question about the duplicate check of jchem.


(The version of Jchem is 3.1.5)


Question:


I attached some structures.(Duplicate.sdf)


Either the PERFECT Search or UpdateHandler.INSERT couldn't check these mols. Please use Jcman to import the SDFile twice.(Don't check All duplicate structures)


You will find Jcman couldn't hit these structures.


(The EXACT Search has a the same problem too)





I would appreciate your early reply.





Best regards


Alex

Dear Alex,





We will check these structures. At the first sight it seems to me that these may represent more than one problem.





Best regards,


Szabolcs

Dear Alex Kuang,





I checked the problems thoroughly.





Structure 2 is already correctly identified as a duplicate in our development system, so it will be OK in JChem 3.2.


In Structure 1 and 3, the internal representation of the 3-membered rings with double bond is having a problem, which will be fixed. In the meantime, you can change the representation as on the attached figures to make duplicate checking recognize the structure.


Finally, at structure 4, the wiggly bond is causing a problem. When you change that to a single bond (which has the same meaning), this will also be recognized at duplicate check. We will investigate this problem further anyway.





Thank you for reporting us these bugs.





Best regards,


Szabolcs

Dear Szabolcs





Thank you for your reply.





I pasted another structure.


It could be hit by UpdateHandler.INSERT,


but couldn't be hit by Exact search.


I would like to know the reason.





About the duplicate check, would you give any suggestion?


(1)The API suggest me use PERFACT search to check the duplcate.


But it seem more bugs than EXACT search.





(2)Do I need use the following methods before run a duplicate check?





setExactChargeMatching


setExactIsotopeMatching


setExactRadicalMatching


setDoubleBondStereoMatchingMode


:


:





(3)I would like to know how JCMAN to check the duplicate?


Does it use UpdateHandler?





Best regards


Alex Kuang

Dear Alex Kuang,

AlexK wrote:

I pasted another structure. It could be hit by UpdateHandler.INSERT, but couldn't be hit by Exact search. I would like to know the reason.

Very strange. It smells like a bug. We will investigate this and get back to you.

AlexK wrote:

(1)The API suggest me use PERFACT search to check the duplcate. But it seem more bugs than EXACT search.

PERFECT match is more rigorous than EXACT. It has to check equality in all features. As it checks more features, it has a higher risk to fail. We will try to make PERFECT search more perfect. :)

AlexK wrote:

(2)Do I need use the following methods before run a duplicate check? setExactChargeMatching setExactIsotopeMatching setExactRadicalMatching setDoubleBondStereoMatchingMode : :

No, this is not necessary, as setting the PERFECT search type all ...Exact... options are set to true.

AlexK wrote:

(3)I would like to know how JCMAN to check the duplicate? Does it use UpdateHandler?

Yes, it does.





Best regards,


Szabolcs

Dear Alex Kuang,





We have found the cause of this mysterious behaviour. It is actually a bug, which is related to some strangely depicted molecules.





One of the stereocenters in your structure is superfluous. (See attached image.) During the search and/or registration process, this superfluous parity info may disappear in the query and/or the database molecule. When the info disappears from only the query, there is no match.





Removing the unnecessary wedge will immediately resolve your problem, and JChem registration and searching will be correct.





We are also planning to solve the perfect search of these superfluous representations.





Best regards,


Szabolcs

FYI, the perfect search bugs above are all fixed now, and the next version (3.2, or the next alpha of it) will be OK with these.





All the best, Szabolcs