Unique charges

The two attached sdf files are for exactly the same structure, only the sequence of atoms in the sdf files is different. However, when I calculate total charges with the option to 'Take resonant structures', I get different charges for the two molecules. For example, the oxygen atom of the amide group gets -0.28 in one case and -0.47 in the other.





The same happens with electronegativities.





Without the option to 'Take resonant structures', properties are unique.





In the GUI window, in one case aromatic bonds are displayed, in the other not, and nitrogen atoms get 5 bonds.





Can anyone find something wrong with my sdf files, or is this a bug?





In order to get unique charges and electronegativities, isn't it safer to choose the option to take resonant structures?





I'm using Marvin Beans installed from marvinbeans-4_1_0-windows.exe on a Windows XP OS.

Hi,








Yes, this is a bug.





If you select the 'resonant structure option' then the weighted sum of the partial charge of the individual resonant structures are calculated.





Without the 'resonant structure option' only the submitted structure is considered.








Both type of charge distribution can be considered as unique.


But of course , including resonant structures result in more life-like charge distribution.














Jozsi

Thanks.





When can a fix be expected?





Meanwhile, could you give an idea about the type of structures for which the charges and oen are expected to be affected by the bug.





Could you confirm that the bug does not affect charges and oen calculated without the option to take resonant structures?

Hi,








This bug will be fixed at this week. I will inform you when we will do a new release (or a bug fix release).





I can not give straightforward descritption about the problematic functional groups.








I can confirm that this bug will attack whenever the 'take


resonant structure' option is selected.


If the 'take resonant structure' option is not selected then this bug does not exist.











Jozsi

I tried the latest test build marvinbeans-4_1_2 and for the two previously mentioned files of the same structure, the total charges calculated with resonant structures are more similar (e.g. for the oxygen of amide they are -0.5177 and -0.5161) but still not the same.





Is this new version supposed to have fixed the bug and such differences are considered to be acceptable, or will it be further improved?

Hi,








...such differences are considered to be acceptable, or will it be further improved?





This difference is not acceptable. We will further improve soon.








Jozsi

Hi,





We have managed to fix this problem.


Now, at least 9-10 digits precision is guaranted for the identical structures.





Jozsi

Hi Jozsi, when is a version incorporating this fix expected?

Hi,








If I know well then a release can be expected on this week. May be it will be shifted a couple days.








Jozsi

Now I tried with Marvin 4.1.3 and I still get cases in which charges are not unique if they are calculated with the 'Take resonant structures' option. I attach two examples. Structures C1.sdf and C2.sdf are exactly the same but get different charges. The same happens with D1.sdf and D2.sdf.

Hi,





The number of the generated major resonance structures is different for the two identical structures.








You can check this with Marvin: Tools ==> Isomers==> Resonance==>Take major contributors








Now, I have fixed this error in the resonance calculation modul.





(See. the attached picture of the calculated charges of C1 and C2.)


This improved Marvin will be released in two weeks.











Jozsi

Jozsi wrote:

The number of the generated major resonance structures is different for the two identical structures.

But the number of major resonance structures for D1 and D2 is the same.

Hi,





Yes, you are right. This was another bug. I fixed already as well.





Thanks for your testing.





Jozsi