rcdk::matches() function bugs

[YANGJJ93MS]

There is an issue for substructure match function. I got true value even if the substructure is not in the query molecule.

Please include a minimal reproducible example (AKA a reprex). If you've never heard of a reprex before, start by reading https://www.tidyverse.org/help/#reprex.

---

There is an issue for substructure match function. I got true value even if the substructure is not in the query molecule.

mol = parse.smiles('CN(C)c1cccc2c(S(=O)(=O)Oc3ccc4c5c3OC3C(=O)CC(O)C6(O)C(C4)N(CC4CC4)CCC536)cccc12')
query =  'CN(C)CCc1ccccc1'
rcdk::matches(query,mol)

Screenshots

"rcdk::matches(query2,mol1)
CN(C)c1cccc2c(S(=O)(=O)Oc3ccc4c5c3OC3C(=O)CC(O)C6(O)C(C4)N(CC4CC4)CCC536)cccc12.match
TRUE"

System (please complete the following information):

• OS: windows 11
• R-4.1.1

[zachcp]

Hi, thanks for your report. So I am less familiar with the SMARTS but here's what I've found:

• rcdk::matches uses SMARTSQueryTool under the hood.
• SMARTSQueryTool is deprecated
• SMARTSPattern is prefferred.
• I am not familiar enough with SMARTS to know if your example is truly expected to be negative. You should probaby confirm on the CDK users mailing list with your specific pattern.
• rcdk should either fix or deprecate this function (@rajarshi would need to weigh in there)

[rajarshi]

If you try it out at https://www.simolecule.com/cdkdepict/depict.html, the supplied SMARTS pattern does match the the molecule. However, I agree that we should fix the function to use SMARTSPattern

…

On Sun, Feb 12, 2023 at 12:19 AM zachcp ***@***.***> wrote: Hi, thanks for your report. So I am less familiar with the SMARTS but here's what I've found: - rcdk::matches uses SMARTSQueryTool <https://github.com/CDK-R/cdkr/blob/master/rcdk/R/matching.R#L16> under the hood. - SMARTSQueryTool is deprecated <https://cdk.github.io/cdk/latest/docs/api/org/openscience/cdk/smiles/smarts/SMARTSQueryTool.html> - SMARTSPattern <https://cdk.github.io/cdk/latest/docs/api/org/openscience/cdk/smarts/SmartsPattern.html> is prefferred. - I am not familiar enough with SMARTS to know if your example is truly expected to be negative. You should probaby confirm on the CDK users mailing list <https://cdk.github.io/cdk/latest/docs/api/org/openscience/cdk/smarts/SmartsPattern.html> with your specific pattern. - rcdk should either fix or deprecate this function ***@***.*** <https://github.com/rajarshi> would need to weigh in there) — Reply to this email directly, view it on GitHub <#136 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAAIMOJR6AXJTU4XQJU5VATWXAUCTANCNFSM6AAAAAARQYH6EY> . You are receiving this because you were mentioned.Message ID: ***@***.***>

-- Rajarshi Guha | http://blog.rguha.net | @rguha <https://twitter.com/rguha>

[YANGJJ93MS]

Dear Rajarshi,

Thank you for your reply!

I found that the rdkit substructer matching fucntion did the same mistake. Please kindly find the picture below:

As a matter of fact, the substructure that I am looking for is an benzene structure with an ammonia side chain, which is totally different from the naphthalene structure. I believed the reason is that the algorithm took the naphthalene structure as an alkyl structure.

Best regards,
Junjie