Absolute configuration switch during 3D generation

Hi,


this is a possible problem regarding FORUM


http://www.chemaxon.com/forum/ftopic3594.html&start=0&postdays=0&postorder=asc&highlight=, if this is the case please confirm. It may also refer to the problem in 3D surface generation in


http://www.chemaxon.com/forum/ftopic4089.html





The calculation from 2D-->3D structures can change the absolute configuration (CIP, R/S) of the molecule in Marvin 5.0.1.





Example molecule SMILES:


COC(=O)[C@H]1C2[C@]3(C)CC[C@@H](O[Si](C)(C)C)[C@@]2(OC3=O)C4CC[C@@H]5C[C@]14CC5=C





Name:





methyl (1R,5R,8R,9S,11S,14R)-11-methyl-6- methylidene-16-oxo-14-[(trimethylsilyl)oxy]-15- oxapentacyclo[9.3.2.1^{5,8}.0^{1,10}.0^{2,8}]heptadecane-9- carboxylate





----------------------------------------------------------------





Tools--> Geometry --> Geometry --> Very strict





3D


[H]C([H])=C1C([H])([H])[C@]23C([H])([H])[C@@]1([H])C([H])([H])C([H])([H])[C@]2([H])[C@@]12OC(=O)[C@@](C([H])([H])[H])(C([H])([H])C([H])([H])[C@@]1([H])O[Si](C([H])([H])[H])(C([H])([H])[H])C([H])([H])[H])[C@@]2([H])[C@]3([H])C(=O)OC([H])([H])[H]








methyl (1S,2S,5R,8R,9S,10R,11S,14R)-11-methyl-6- methylidene-16-oxo-14-[(trimethylsilyl)oxy]-15- oxapentacyclo[9.3.2.1^{5,8}.0^{1,10}.0^{2,8}]heptadecane-9- carboxylate








2D


COC(=O)[C@H]1C2[C@]3(C)CC[C@@H](O[Si](C)(C)C)[C@@]2(OC3=O)C2CC[C@@H]3C[C@]12CC3=C








methyl (1R,5R,8R,9S,11S,14R)-11-methyl-6- methylidene-16-oxo-14-[(trimethylsilyl)oxy]-15- oxapentacyclo[9.3.2.1^{5,8}.0^{1,10}.0^{2,8}]heptadecane-9- carboxylate








----------------------------------------------------------------


So the problem in case of drug research,


it undergoes 2D to 3D conversion (already erroneous) and somebody cleans it back to 2D it would be a totally different molecule. I also observed this during conformer generation.


I would have no problem if it states (absolute) configuration


changed, during a calculation, although this may be a big problem for others. But I would have rather a info message,


instead of a wrong result.





I remember that I had similar problems during conformer generation (switch of R/S config), do you know if this problem was solved or is it still under development?








----------------





I was looking at the molecule itself a bit longer, maybe the configuration


itself is "impossible" although its a correct chemical graph,


so the algorithm tries to clean that up and changes to a configuration with a higher probability? But then there should be at least a warning or small info somewhere.





Thank you


Tobias

Dear Tobias,





We have introduced major Clean3D modifications in Marvin version 5.1, however it seems that the above mentioned structure is handled incorrectly. We will notify You in the forum about the findings.





A problem (which is discussed in http://www.chemaxon.com/forum/ftopic3594.html&start=0&postdays=0&postorder=asc&highlight= ) is fixed in this version, so the corruption of *already specified* chirality requirements seems to be eliminated.

Quote:

So the problem in case of drug research, it undergoes 2D to 3D conversion (already erroneous) and somebody cleans it back to 2D it would be a totally different molecule.

Quote:

I was looking at the molecule itself a bit longer, maybe the configuration itself is "impossible" although its a correct chemical graph, so the algorithm tries to clean that up and changes to a configuration with a higher probability?

After a 3D conversion every stereocenter will realize exactly one possible configuration in the 3D structure. If the chirality specified for any of the centers then it will be fulfilled or (if the structure is not possible to realize) an error message will be given. So no configuration is changed during normal operation. However, after a conversion back to 2D all of the stereocenters will have specified chirality, so the initial and the resulting 2D structures can be different.





If you have any further questions, please do not hesitate to ask them.





regards,


Gabor

Dear Tobias,





It seems that the above mentioned example SMILES (attached as in.smi) represents a chemically unfeasible configuration: i think no 3D low energy conformer exists having this configuration. Consider the attached in.core.smi which is the problematic" region of the original structure.





I can confirm that the 3D coordinate generation process returns a 3D structure with one corrupted chirality; it should give an error message instead. We will debug this and will notify You.





Regards,


Gabor