compounds not found by JCHEM cartridge

Compounds having the scaffold on the right are not found when searching with the query depicted on the left. We're using Marvin 5.2.1_1 and the latest JCHEM cartridge.

Thanks.

Hi Sorel,

The single bonds in the query between the two rings and going towards the terminal N prevent matching. (Explicit single bonds in the query don't match aromatic bonds, except if they can be aromatized. This means that they must be part of an aromatic ring's Kekulé representation.)

 

There are currently two solutions:

1. Instead of single bonds, "single or aromatic" or "any" bond types should be used.

2. Vague bond level 3 should be set.

 

Furthermore, for version 5.3 we were planning a new vague bond option that solves aromatic matching of one-atom single bond ligands of aromatic rings. (Like the -N ligand in your example query.) We will investigate to extend that method for single bonds between (potential) aromatic rings. (Like the other single bond in your query.)

 

The plan is that this new vague bond option will be default. (And will also involve the current default vague bond level 1.)

 

Best regards,

Szabolcs

Hi Szabolcs,

Thanks for a quick replay. This makes sense although in our experience increasing the vague bond level decreases somewhat the searching performance. I look forward to testing the 5.3 version.

Talk to you soon,

Sorel.

 

 

sorelmuresanAZ wrote:

This makes sense although in our experience increasing the vague bond level decreases somewhat the searching performance.

 

That is true. I just played around with the query and the search options, and it seems that the single/aromatic solution is much more efficient than vague bond level 3.

I also found that if the query is drawn with aromatic bonds (like: [#7]-c1nccnc1-[#6]~1~[#7]~[#6]~[#6]~[#7]~1 ) , performance of vague bond level 3 goes up to the same level.

We will check if we can improve performance for the Kekulé query as well.

 

sorelmuresanAZ wrote:

I look forward to testing the 5.3 version.

I forgot to mention: version 5.3 is scheduled to this autumn.

Talk soon at the UGM,

Szabolcs

Hi Sorel,

Just for the record: JChem 5.3 has long been released, and we solved all of the above issues:

- By default (vague bond level 1 and up), bridging single bonds between potentially aromatic rings will also match bridging aromatic rings

- And we also increased performance of these queries with vague bond level 3. (We exploited the fact that the 6-member ring cannot be aliphatic.) With this, now vague bond level 3 is basically the same performance as vague bond level 1 for this query on the 260K NCI database.

 

Best regards,

Szabolcs