wavy bonds

Hello,





We had a discussion at the UGM about the meaning of wavy bonds. From the IUPAC Basic Terminology of Stereochemistry recommendations from 1996:

Quote:

If stereochemistry is unknown this can be indicated explicitly by a wavy line.

Quote:

A wavy line can be used to indicate either that the stereochemistry is unknown (7), but only one form is present, or, if explained in the text, that both isomers are present and will be defined when required. If it is intended not to show any stereochemistry it is best to only use plain lines for all bonds.

It appears that even IUPAC acknowledges the existence of two conventional meanings of the wavy bond. So I think you should provide users the option to use one or the other when matching structures.





-- Bob

(I moved this topic to the search forum.)





OK, we will consider this recommendation. You need it mainly in MolSearch, don't you?

We need it in whatever method is use to compare two structures, but I imagine others would need it in a database search.

We are planning to introduce an option called isWigglyRacemic(). Its default would be false, which corresponds to the current behaviour. However, when it is set to true, wiggly bonds would work exactly as a plain bond.





Bob, is this acceptable for you? This is what I can read from the citations above, but Gyuri says you had a different opinion.

I remember, that you proposed, that wiggly wedge should match another wiggly wedge only, while a plain single bond matches up and down wedges as well.

My opinion is that in one option, wavy bonds should behave just like bold or hashed bonds. That is, a wavy bond should match only to a wavy bond, but a plain bond should continue to match to any bond.





I would call the parameter isWavyMixture(). If it is true, then a wavy bond would mean that there is a mixture of stereosiomers. Plain bond would be unspecified -- it could be just one isomer or a mixture. If isWavyMixture() is false, then a wavy bond would mean one or the other isomer, but we don't know which -- the MDL definition.

Hi Bob,





OK, that also makes sense. We will make this behaviour then for the "true" option value.





This topic now reminds me the enhanced stereo representation:





http://www.chemaxon.com/jchem/doc/user/Query.html#mdl_enhanced_stereo





You can also put it that way: isWavyMixture=FALSE corresponds an OR stereo center, and isWavyMixture=TRUE to an AND stereo center.

Yes, that's a brilliant way of describing the difference, with one slight modification: I believe the MDL convention (your current convention, and the presumed default) is that wavy bond means XOR, not inclusive OR.





In fact, I think a better name for the flag would be wavyBondMeans(), and it could be set to "XOR" (default) or "AND". I suppose you could also add an inclusive "OR" option if you really want to go all out; in that case, the wavy bond would be identical in meaning to the plain bond.

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

Changes in JChem (Look for "comparators". One of the examples offers a solution for wavy bond matching.)


Apidoc for MolComparator

Tamás