Marvin Sketch question

We are BindingDB (www. bindingdb.org) data curation team. We have a question about specification of stereochemistry. At the moment, we express uncertain stereochemistry at a chiral center by using all straight (non-wedge) bonds. However, the reality is that there are two different ways in which the stereochemistry of a chiral center can be undefined in an article. One is that the authors don't tell us anything at all about the stereochemistry. The second case is that the authors state specifically that they are working with a racemic mixture. It would be nice if Marvin/JChem provided two different types of bond to capture this difference. One possibility would be to distinguish a racemate by putting a little "R" next to a bond.





Or do you already have a way of making this distinction?





Thank you in advance,

If you are concerned about only a single stereocenter, then the wavy bond can be used to denote a mixture of stereoisomers. (Although the wavy bond is used by some to designate a center with fixed but unknown configuration.)





See also this discussion.

Hi,





Bob have given a nice answer, but you can use MDL enhanced stereochemical representation. See:


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





Which is also cited by Bob:


http://www.chemaxon.com/marvin/help/sci/stereo-doc.html





Thank you Bob for the answer!





Andras

Hm, it seems the wiggly bond usefully conveys uncertainty about the stereocenter, but I'm not sure how to distinguish between the following two cases:





1. article gives no information about the stereochemistry, and BindingDB wants to preserve this ambiguity.





2. article explicitly states the measurement is for a mixture of both stereoisomers, and BindingDB wants to convey this information.





How about if we use a plain (non-wedge) single bond for case 1, and a wiggly bond for case 2?

Your solution works well if there is only one stereocenter in the compound.





However, if there is more than one stereocenter, it does not necessarily work, because the authors might designate a particular relative stereochemistry without denoting the absolute stereochemistry. If your compound has more than one stereocenter, and you want to indicate that it is a diastereomerically pure racemic mixture, you need either to use the MDL enhanced stereochemical representation or chiral flag mentioned by Andras, or you need to include both enantiomers explicitly in the molecular definition.

Thanks. We are considering switching to MDL enhanced stereochem. This leads to two questions:





1. If our database includes MDL enhanced stereo specifications, but the user sets up a query from marvin using traditional notation (no "abs", "&1", "or1", etc.), how is the query processed?





2. We have about 18,000 compounds in the traditional format.


a. What happens if the user queries these using MDL enhanced stereo?


b. Is there a tool to convert from traditional stereochem to MDL enahnced?

1 & 2a. I've set up this JSP page that allows you to see how JChem behaves given any single query and target. It currently uses JChem 3.2.12, not the very recently released JChem 5.0. You should also take a look at the JChem query guide.





2b. I'll let ChemAxon answer this one.

Quote:

how is the query processed?

See "Matching rules of the enhanced representation"


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





In MDL's enhanced stereo representation all stereo center atoms are labeled with one of the following:





1. ABS


2. ORn


3. ANDn

Quote:

2. We have about 18,000 compounds in the traditional format. a. What happens if the user queries these using MDL enhanced stereo? b. Is there a tool to convert from traditional stereochem to MDL enahnced?

Depending on what do you expect from the traditional format you should convert the molecules using API functions. As the expectations are different there is no convert tool for it.





At first you should decide how do you interpret your traditional data.


You may read the section at:


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


Considering "MDL old stereo representation (chiral flag)" versus "MDL Enhanced Stereo Representation"





Andras

You can also take a look at and play with the "Substructure Search example" using the latest JChem and Marvin at:


http://www.chemaxon.com/jchem/examples/sss/index.jsp

I have inserted a few example records in this example for table editexample:


http://www.chemaxon.com/jchem/examples/db_search





You can retrieve them by the query CC(Cl)CCCl . I used the same example to explore the search behaviour.

bindingdb wrote:

Thanks. We are considering switching to MDL enhanced stereochem. This leads to two questions: 1. If our database includes MDL enhanced stereo specifications, but the user sets up a query from marvin using traditional notation (no "abs", "&1", "or1", etc.), how is the query processed?

If there is no stereo information specified in the query (plain single bonds): all records will be returned.





If the query contains a stereo center without labels, it will only return records labeled with abs. (By default, it also means that unlabeled stereo centers are returned, because the "Assume absolute stereo flag" is on by default for table creation. See:


http://www.chemaxon.com/jchem/doc/admin/#create )





However, there are search options that can change this behaviour. For example, setStereoSearch(false) will return all records again. (Maybe you should provide that search option for your users. See the advanced options tab on the example applications' Query page.)





http://www.chemaxon.com/jchem/doc/user/Query.html#optionsSummary width="90%" cellspacing="0" cellpadding="3" border="0" align="center"> bindingdb wrote: 2. We have about 18,000 compounds in the traditional format.


a. What happens if the user queries these using MDL enhanced stereo? Because of the above mentioned absolute stereo flag, all stereocenters in your old compounds collection will be assumed to have abs labels. The search will behave accordingly.

bindingdb wrote:

b. Is there a tool to convert from traditional stereochem to MDL enahnced?

Unfortunately, there isn't yet, but we are planning a new Standardizer action. However, it will be your choice what label to assign to the ambiguous cases. (Ambiguous case: no chiral flag may mean AND or OR configuration.)





The absoluteStereo action may also be useful, but the above mentioned table option handles that case correctly.