Missing values when importing data

Hello dear Chemaxon community members,

I want to create a small table for superstructure search. My point is to be able to pass SQL queries for preselection before the actual superstructure search. Thus, I have created a table of type "Query structures", and added three columns named nr_heteorings, largest_rings_ystem_size, and nr_of_fused_rings. The names described exactly the content of these fields and I am using chemaxon library functions to fill up the fields.

Those three values will be used with SQL statements and compared btween the query structure and the one in the table. Only the ones returned will be used for superstructure search. The table is built by importing structural motifs using jcman.

Surpisingly, none of the three values was assigned for most of the structural patterns I imported. A small subset of the table is attached here as a sdf file.1-Phenylpropene is an exemple of a structure that was not assigned any of the three features. Surprisingly to me, the smarts attribute was NULL. This lead me to use "molconvert smarts " to convert 1-Phenylpropene.mol  to a smarts expression. This operation was succesfull ([#6]!@\C!@=C!@\[C;R1]1=[C;R1][C;R1]=[C;R1][C;R1]=[C;R1]1) .

(1) What could cause jcman not to assign a smarts expression, during the "import" process while molconvert is succesful at it?

While looking at the structural patterns which were not assigned any of the three aforementioned values (nr_heteorings, largest_rings_ystem_size, and nr_of_fused_rings), I realized that many of them were in fact assigned proper smarts expression. This lead to the present of a smarts expression does not implies the succesful assignment/computation of these features. Looking at the fields formula, and sortable formula, I came to the same conclusion.

Could you please help me solve this issue?

Thanks,

 

MrYan

 

Hello MrYan,

Regarding your question about empty chemical term fields, we have to tell you that our chemical terms calculator doesn't calculate all types of chemical term values for molecules with query properties (query atoms, query bonds, ...) only for specific molecules independently of the possibility of smarts representation.

As for the empty cd_smarts field, in the latest released JChem version (5.12) we have decreased the number of null values in the cd_smarts field of  query tables, but it didn't succed yet in case of every query properties assigned to the structures. It depends not only on the operation of molconvert, but we have to ensure that the search processes using this field run trustworthily. We are working on this issue.

Please don't hesitate to contact us if you have further questions.

Best regards,

Krisztina

 

Hi Kriztina,

 

Have you guys found a solution to this problem? I am using the 6.1.2 version and I still get NULL smarts when I import structures into the jchem table using jcman, although molconvert gives me non-empty smarts.

What could be the cause for that?

Thanks for replying.

MrYan

Hi MrYan,

We didn't change the logic of the chemical terms calculator in the meantime. It doesn't calculate all types of chemical terms for query structures (query atoms/bonds, query atom/bond properties, R-group structures, ...).

However, we have improved the smarts/cxsmarts import function. Now I found that the cd_smarts field was not NULL with JChem v6.1.x for the structures in small_test.sdf you attached to this topic.

Could you send us examples of NULL cd_smarts when molconvert generated non-empty smarts?

Thank you,

Krisztina

Hi Kriztina,

 

Here is another file. I am using version 6.1.3.

Another problem is that using webservice, I cannot even import all the structure names. Here is the syntax I use (with ruby on rails):

options['data'] = [ { "structure" => mol, 'additionalData' => { 'name' => name } } ]

RestClient.post create_url, options.to_json, content_type: :json, accept: :json

where "create_url" is the URL of the table, and "name" is the name of the structure file to be imported. I would expect at least attributes such as "name" that do not need to be computed but only "copied and inserted" be all included in the database. however, some names are not. I could not see what is the cause.

Thank you for your help.

MrYan

Hi MrYan,

I tried to reproduce the missing name import from test_Nov28.sdf file into a Query structures type JChem table after modifying the table by adding a column for the name of the structures. All the names were imported successfully. Therefore, I move your question to our Web Services forum topics. They will answer you soon.

Best regards,

Krisztina

Hi Kriztina,

 

Thanks for your reply. I hope to have an answer soon.

Thanks again.

MrYan