Inconsistent results with Stereoisomer generation

User 81a38f9467

08-12-2016 23:34:08

I am generating stereoisomers using the Stereoisomers node in KNIME, using both the v501 nodes (JChem 15.12.7) and the v505 nodes (JChem 16.11.21).  The node is configurated to generate tetrahedral stereoisomers, Protect tetrahedral stereocenters, and Filter invalid 3D structures.   My two input structures are two depictions of 3-(3-iodoadamantan-1-yl)oxolane and, since these two represent the same molecule, I would expect them to produce identical outputs.


What I observe is that the two depictions produce different outputs.  Both inputs should produce two valid output structures.  When I use the v501 nodes, Example_1 generates 2 valid isomers and Example_2 generates 6 valid isomers.  However, when I use the v0505 nodes, Example_1 generates 6 valid isomers and Example_2 generates 2 valid isomers. 


Question 1:  Why do the two structures generate different results?


Question 2: Why are the results reversed when I switch from JChem15.12.7 to JChem16.11.21 ?


Please let me know.


Thanks,


Tim


 


ChemAxon d51151248d

15-12-2016 16:15:42

Hi Tim, 


Sorry for the late response. We are investigating the issue and working on the solution. We will notify you as soon as the fix is ready.


Best regards, 


Daniel


 

ChemAxon 5fc3e8d7d0

23-12-2016 14:31:20

Hi Tim,


The mentioned issue was fixed and the fix will be released in JChem version 16.12.26
Thanks for your remarks. 


Best regards,
Laszlo

User 81a38f9467

27-01-2017 01:01:29

Is this really fixed?


Using JChem 17.1.2 I am still seeing two isomers for Example_1 and six isomers for Example_2.   I don't understand why the results for the two structures are different.   


The correct answer for both examples is two isomers.





ChemAxon d51151248d

27-01-2017 15:47:24

Dear Tim, 


I couldn't reproduce the issue in MarvinSketch and cxcalc (coming from the JChem bundle) v. 17.1.2, so the issue is fixed in JChem. Please see the attached screenshot for the test results. 


However, the fix needs to be integrated in KNIME. The new KNIME release that will have this fix will be available soon, most probably next week. 


Best regards, 


Daniel

User 81a38f9467

31-01-2017 21:54:05

Thank you for your response.  I now agree that the issue is resolved in 17.1.2.


Did the method for filtering invalid 3D stereoisomers change?   Can you explain how that filter works?


In the past, I would generate all stereoisomers with Protect Tetrahedral Stereocenters, take the resulting structures and calculate the Dreiding energy, then remove structures where the Dreiding energy calculation failed --- this process output only valid 3D stereoisomers.  But with more recent versions of JChem this process is no longer valid --- the Dreiding energy calculation returns values for some of the invalid structures.  


Any insight would be appreciated.


Thanks,


Tim

ChemAxon d51151248d

06-02-2017 13:17:42

Hi Tim, 


Yes, the 3D filtering method has changed recently. In the newer version there is an additional filtering step after the 3D structure generation that checks different properties, for example keeping the R/S stereo configurations in the generated 3D molecule the same as in the 2D structure.


Best regards, 


Daniel