Why a reaction could not be 'exact' search out

Hi,





A few reactions can be searched out using cd_id, but when I copied its structrue from MarvinView window and past it in the Marvin of query.jsp, use 'exact' search, I got 'no hit'. I also tried to copy it into Marvin of edit.jsp and 'insert' it into database, it really would be inserted again though the reaction has existed in the database table (before inserting, there's a 'perfect' searching to avoid such issue in the update.jsp, but the reaction seems also can not be 'perfect' search out, see the following error message:


java.lang.ArrayIndexOutOfBoundsException: 0


at chemaxon.jchem.db.JChemSearch.getResult(JChemSearch.java:2005)





The attached *.mrv is such a reaction.





Thanks,


Xu

Hi,





We will investigate the problem and return to you soon.


Thanks for the patience,


Robert

Hi,





There is a small bug in the hash code calculation for reactions. We will fix it soon.


Bye,


Robert

Hi Xu,





The new version 5.1.3 released last week contains the fix of this bug.


The correction will require the regeneration of reaction tables.


Thanks for reporting this bug.





Robert

Quote:

Hi Robert, I update the jchem to 5.1.4, and still found this bug. I found if the structure talbe created with ' --t:any', there's such a bug, and the searching speed is very slow(waiting.....checking.....). All reactions with contracted group in this table could not be perfect search out after inserted them. But if the structure table created with '--t:reactions', there'll be no such bug. However I would like to use  structure talbe with 'any' property to store reactions, a few of which may be in markush format.  Thanks, Xu

We would like to understand the problem. Could you please attach some example image of those reactions?

Hi,





One example is the attachment in my first post for this topic.





Thanks,





Xu

Hi Xu,





We thought you were using reaction tables, and fixed the perfect/exact search problem for it.





Now I checked any tables and could reproduce problem. We will notify you when it is fixed.





Bye,





Robert

Hi Xu,





We fixed the bug concerning searching of reactions with sgroups in any table.





Earlier we fixed a bug regarding bad hash code calculation. This time the problem was caused by inconsitent handling of sgroups.





We are sorry for the inconveniences and that this sgroup bug remained hidden the last time (for reaction tables).





The bugfix will appear with version 5.2 which is due to be released in the first half of march.





Bye,





Robert