ChemAxon Forum Archive » Storage & search: JChem Base and JChem Oracle Cartridge

wavy bond matching

User 870ab5b546

07-09-2006 13:01:11

At the UGM, we discussed adding an option to MolSearch to make a wavy bond mean UP && DOWN, rather than the current meaning of UP || DOWN && !(UP && DOWN). Any progress on adding this option?

ChemAxon a3d59b832c

07-09-2006 13:23:30

This option is scheduled to appear in JChem 3.3.





In the meantime, you can use the enhanced stereo configuration, as we discussed in the other forum topic here:





http://www.chemaxon.com/forum/ftopic1546.html&highlight=wavy

User 870ab5b546

07-09-2006 14:15:58

I don't think the MDL enhanced stereo configuration will serve our purposes. We need to match a wavy bond to a wavy bond but NOT to an up or down bond. How does MDL handle a stereocenter with a wavy bond that has been labelled OR or AND? If we change the wavy bond to, say, an UP bond and label it AND, it will match to another UP bond, which is not what we want.





When might we expect JChem 3.3? Days, weeks, months?

ChemAxon 9c0afc9aaf

07-09-2006 15:10:07

Hi,
Quote:
We need to match a wavy bond to a wavy bond but NOT to an up or down bond.
Using ExactStereoMatching may just solve this problem, if this is suitable for you:





http://www.chemaxon.com/jchem/doc/api/chemaxon/sss/search/MolSearch.html#setExactStereoMatching(boolean)
Quote:
When might we expect JChem 3.3? Days, weeks, months?
Several months.


(Our next major release, 3.2 will just come out soon, and 3.3. is the major release after that)





Best regards,





Szilard

User 870ab5b546

07-09-2006 15:38:54

I don't think so, because we need a regular bond in the query to match to anything in the target.





Here's our desired behavior:





Code:



                        target


query         up   down   wavy   unspec


up            y    n      n      n


down          n    y      n      n


wavy          n    n      y      n


unspec        y    y      y      y

ChemAxon a3d59b832c

07-09-2006 19:35:02

OK, we may be able to deliver it in one of the patch releases, before 3.3.





So the other options for the simulation of enhanced stereo-like behaviour (AND or XOR or OR) as discussed in the other topic are not important any more?

User 870ab5b546

07-09-2006 21:14:51

Your industry clients want XOR. We need AND. I don't know anyone who would want inclusive OR, but I suppose there's no reason not to offer it.





The AND option should deliver the behavior I describe above.

ChemAxon a3d59b832c

08-09-2006 07:32:55

OK, Thanks.

ChemAxon a9ded07333

16-01-2008 14:12:43

From JChem 5.0 we are supporting custom MolComparators. More information is available atTamás

User 870ab5b546

16-01-2008 15:05:33

Thanks, this feature looks like it might allow us to cut out a lot of workaround code.

User 870ab5b546

17-01-2008 16:51:55

Hm, something is not working like I expect.





I used your StereoANDComparator class unchanged. I modified ExampleForMolComparators to do more comparisons. Here's the output:





Code:
********************************





* Testing StereoAND comparator *





********************************





mol1 = ClC1CCCCC1Br





mol2 = ClC1CCCC[C@@H]1Br |r|





mol3 = ClC1CCCCC1Br |w:6.7|





Without comparator


Target: mol1, query: mol2


1. search:


        No hits


Target: mol1, query: mol3


2. search:


        No hits


Target: mol2, query: mol3


3. search:


        Hit 1:  0 1 2 3 4 5 6 7


Target: mol2, query: mol1


4. search:


        Hit 1:  0 1 2 3 4 5 6 7


Target: mol3, query: mol1


5. search:


        Hit 1:  0 1 2 3 4 5 6 7


Target: mol3, query: mol2


6. search:


        No hits





Using comparator


Target: mol1, query: mol2


7. search:


        No hits


Target: mol1, query: mol3


8. search:


        No hits


Target: mol2, query: mol3


9. search:


        No hits


Target: mol2, query: mol1


10. search:


        No hits


Target: mol3, query: mol1


11. search:


        No hits


Target: mol3, query: mol2


12. search:


        No hits






Now, if I'm not mistaken, searches 10 and 11 (query mol1, with unspecified parity) should return hits, but they don't. What am I missing?





I'm attaching the modified ExampleForMolComparators.java for your perusal.

ChemAxon a9ded07333

18-01-2008 12:00:42

You are right, the mentioned searches give wrong result.


There is a wrong constant in StereoANDComparator.java: instead of
Code:
query.getLocalParity(queryAtom) == PARITY_UNSPEC
one should write
Code:
query.getLocalParity(queryAtom) == 0



I corrected the example on the ChemAxon site as well.

User 870ab5b546

18-01-2008 12:24:27

Thanks, that does the trick.





This technique will be *very* useful for us.