Search in Markush Compounds

In ChemAxon standard presentation I was attracted by an intention to offer in the near future "Search in Markush compounds". As I am highly interested in this problem I would like to learn more about the concept and expectations.

I would rather say this is a long term goal.





Expectations: At first we plan to allow R-groups, multiple groups with variable repetition, link atoms in the target. (Multiple groups with fixed number of repetition are already supported.)





Later some conditions will also be used (e.g. ones that can be used now for R-group queries), and other features based on user request and feedback.





You can find much information about Markush searching in JCICS, where the Sheffield group published several articles.





All the best,


Szabolcs

I was looking for a database engine that would allow generic substitutents (Rs) as a part of the structure yet permitting (through a link or separate field perhaps) the explanation what the particular R stands for. The substructure search should then reveal respective core structures but the information about proper Rs would be available for each record.





I understand that you aim the R-groups development as a part of query language only ?





Jiri

Jiri wrote:

I was looking for a database engine that would allow generic substitutents (Rs) as a part of the structure yet permitting (through a link or separate field perhaps) the explanation what the particular R stands for. The substructure search should then reveal respective core structures but the information about proper Rs would be available for each record.

You can already include R-atoms in the structure inserted into JChem, so the storage of the core structure (or scaffold) is available. You can add custom columns to JChem tables where you are free to store information about the substituents. (However, these custom columns will not be considered during substructure search.)





In the future we intend to store and search the whole generic structure (including R-group definitions) in the database.

Jiri wrote:

I understand that you aim the R-groups development as a part of query language only ?

On the query side R-groups are already available for a long time, see the relevant section of the JChem Query Guide:





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

>>You can already include R-atoms in the structure inserted into JChem, so the storage of the core structure (or scaffold) is available.





That sounds great. But can I also make the pseudoatoms (e.g. R1, R2, ...)part of a substructure search query ? For instance to look for substructure defined as pyridine 2-substituted with R1 and 4-substituted with R2 ?





>> In the future we intend to store and search the whole generic structure (including R-group definitions) in the database.





That would be perfect match for what I am looking for (handling large libraries mostly defined as Markush structures). Could you be more specific how the intented search engine would work ?

jirikrechl wrote:

>>You can already include R-atoms in the structure inserted into JChem, so the storage of the core structure (or scaffold) is available. That sounds great. But can I also make the pseudoatoms (e.g. R1, R2, ...)part of a substructure search query ? For instance to look for substructure defined as pyridine 2-substituted with R1 and 4-substituted with R2 ?

Yes. I inserted the attached structure into our demo database:


http://www.jchem.com/examples/jsp1_x/initsearch.jsp


(Select table "editexample".)





However, the search for this query will also match many other structures, because undefined R-groups will match every atom, including Hydrogens (explicit & implicit as well).





If it helps, we could implement an option that R-atoms should only match other R-atoms with the same number.

jirikrechl wrote:

>> In the future we intend to store and search the whole generic structure (including R-group definitions) in the database. That would be perfect match for what I am looking for (handling large libraries mostly defined as Markush structures). Could you be more specific how the intented search engine would work ?

We haven't decided the details yet. One option would be enumeration, but this is not always possible.





We definitely would like to find specific substructures of a generic structure, even if it spans across the scaffold and R-group definitions as well.





Best regards,


Szabolcs

Hello Jiri,





We have implemented a prototype for Markush searching, which is already available to try in alpha version. This presentation at the recent User group meeting tells about some details:


http://www.chemaxon.com/forum/viewpost6503.html#6503





We continue to work on the project.





Szabolcs