Forbidding wedge bonds off of planar (aromatic) atoms

We have a problem (partially but not entirely social in nature) that we're hoping there's a technical solution to.





A certain number of structures have been entered into our JChem database that match the template shown on the left of the attached file. This is a problem for a couple of reasons:

1. Everybody immediately knows what it means, but roughly half of them know that it means something other than what the other half think.
2. We export structures from JChem to a couple of other programs (ACD and ISIS) that have their own unique interpretations, at least one of which is exactly the opposite of JChem's (JChem seems to view the two structures shown below as equivalent, but ACD does not).

To get away from this problem, we need to be able to do two things: first, we have to detect when such structures are being registered, intercept them, and politely require the user to change the structure to an unambiguous form; second, we need to find all the existing structures that match {wedge bond with origin on aromatic atom}, and get them fixed.





The first seems like it would be easy enough, I'm just not sure quite where to start looking. The second, so far, has totally flummoxed me, since every time I search for something resembling that pattern, JChem kindly reinterprets it for me, and finds every structure with stereocenter adjacent to an aromatic atom (of which there are inconveniently many). Can anyone advise me what we can do, or where I should be looking for more information about this?





Edit: and of course I forgot the critical information... we're using Marvin 4.0 and JChem 3.1.4.

Hi Ben,





Sorry for the delay in the answer, we were discussing the problem here intensively. Searching for these structures does not work because our search algorithm does not directly operate with the wedges, but the chirality(parity) information resulting from these. (This is what ensures that chirality is used properly in 0D(smiles), 2D and even 3D structures in every combination as queries and targets.)





We will implement a new function to identify these kind of non-preferred configurations. Next week we will put here some Java code for this new function, which you will be able to build into your registration process and to check your existing structures. Later this code will be part of the Marvin and JChem APIs as well.





Is this solution suitable for you?





Some more, related information:

• There is one useful function in Marvin Sketch related to this problem: Menu item "Edit > Clean > 2D > Wedge bonds" will exactly convert your example molecule into the correct representation. (This was already in Marvin 4.0.)
• From Marvin 4.1 API we already have something related:MolAtom.isAmbigousStereo(). This identifies when chirality cannot be interpreted at all, but certainly it is a useful check at registration, as well as MolAtom.hasValenceError().

Best regards,


Szabolcs

Thanks for the reply! And my apologies in turn for the delay on this end (we had a three-day weekend). I think the solution you propose will work admirably.





And thank you especially for pointing out that cleaning option--I believe I'd seen it previously when looking for something else, but hadn't realized what it did. I suspect it will be very helpful.





Regards,





Ben.

Hi Ben,





Attached is the implementation, with some test data. It works with Marvin 4.0. From the next major Marvin version (4.2), this method will be part of the API. (We are still pondering which class to put into.)





Regards,


Szabolcs

Hi Ben,





In Marvin 4.2 the chemaxon/marvin/util/CleanUtil.class will contain the isNonPreferredStereoAtom(m) method which will return a boolean array corrsponding to the atoms.





Andras

Great--thank you all very much!