Double and Triple Bonds in MarvinSpace

Hi,


while editing a list of molecule views of all aminoacids (using the templates) I got problems with the presentation of the aromatic rings of tryptophan, phenylalanin etc: There are no double bonds in the MarvinSpace presentation. Only the carbonyl double bond is shown in these amino acids.





So I produced the test molecule shown in the screenshot.





Only the vinyl double bond is shown, the aromatic double bonds and the triple bond are lacking. (And the H-Atom at the sp-Carbon has a bad position).





The molecule is transferred either by "View-Open MarvinSpace", or by Copy/Paste to MarvinSpace or by opening the mrv-file of the structure: Always the same picture.





In MarvinView the bonds are all OK.





How to get the right 3D presentation?





Regards, Hans-Ulrich

Dear Hans-Ulrich,





Thanks for the bug report.

Quote:

There are no double bonds in the MarvinSpace presentation.

There actually is double bond in MarvinSpace. However triple and aromatic bonds are not distinguished from single bonds now. Since structure is aromatized on opening in mspace the double bonds are tricky (but possible) to display:





I recommend to open in mspace the saved structure, right click on the right tree control over "Ligand" and select "Open in MarvinSketch", then dearomatize the structure in the sketch using structure - aromatization - dearomatize.

Quote:

(And the H-Atom at the sp-Carbon has a bad position)

It seems that the explicit H atoms are added after Clean3D. This separate hydrogenization method seems bogus, we will fix it asap. An optimization on the hydrogenized structure (or invoking Clean3D on it) should work. (We planning to implement the partial optimization of the added H atoms in the future)





It is clear that mspace could be improved; i hope that we will be able to allocate developer efforts on this.





Regards,


Gabor

Dear Gabor,


thank you for your quick answer. I immediately used the workaround you have explained:





1) Opening the molecule in MarvinSketch and adding the hydrogens.


2) Saving it as *.mrv


3) Opening in MarvinSpace, adding Display-DrawType-Ligand-Hydrogens


4) Right click on the right tree control over the "yellow Ligand" and "Open in MarvinSketch"


5) Use Structure-Aromatization-Dearomatize


and immediately the double bonds are shown in the MSpace window !





And there are now two ligands in the right control tree (see attached screenshot),


and you can click back and force between these two ligands = molecule views.





First dearomatizing the structure in MSketch, saving and then opening in MSpace does not work.


You have to do the dearomatization "inside the ligand" in MSpace.





So now it is more important to implement the display of triple bonds.





MarvinSpace is a super program with many many features!





Regards, Hans-Ulrich

Dear Gabor,


evaluating the options in MarvinSketch I found the results shown in the (edited) screenshots below.





The option "View - Open Marvin View 3D" works fine both for the dearomatized molecule left, as for the aromatized molecule right.





(The rotation of the vinyl-group obviously comes from the intermediate 3D clean up, but I cannot see why the new conformation is better. A conformer calculation gives 22.48 kcal/mol for the upper conformation in the MarvinSketch windows, and 22.51 kcal/mol for the lower conformation in the MarvinView windows, independant of the aromatization/dearomatization.)





But the option "View - Open MarvinSpace" produces inacceptable results. And it is impossible to edit the representation as discussed above in the "true" MarvinSpace program (when one tranfers the molecule with copy/paste from MarvinSketch to MarvinSpace or when one opens the file in MarvinSpace).





The MarvinSketch option "View - Open MarvinSpace" should be deleted.


In this present form it is impractical to use it.





To say it once more: MarvinSpace itself is a super tool to produce images of molecules, only the triple bonds are lacking in MarvinSpace.





Regards, Hans-Ulrich

Dear Hans-Ulrich,

Quote:

The option "View - Open Marvin View 3D" works fine both for the dearomatized molecule left, as for the aromatized molecule right.

Since temporal hydrogenization and aromatization are parts of the standardization step invoked in the Clean3D process, the two representation should give equal results.

Quote:

The rotation of the vinyl-group obviously comes from the intermediate 3D clean up, but I cannot see why the new conformation is better.

It is not necessarily better. The Clean3D invoked on opening mview3D should produce a reasonable 3D structure (of course in the scope of the applied force field) and not an estimate to the best, lowest energy one.

Quote:

But the option "View - Open MarvinSpace" produces inacceptable results. And it is impossible to edit the representation as discussed above in the "true" MarvinSpace program (when one tranfers the molecule with copy/paste from MarvinSketch to MarvinSpace or when one opens the file in MarvinSpace).

I would like to recall our recent discussion in http://www.chemaxon.com/forum/viewpost18732.html&highlight=#18732 where i intended to make clear that we are planning to make a full featured marvin space interface available for the above mentioned "open marvin space" function.

Quote:

The MarvinSketch option "View - Open MarvinSpace" should be deleted. In this present form it is impractical to use it.

I would like to argue this opinion. MarvinSpace visualization even with the current limitations can come handy for medium sized structures (like C60). Also please consider the "stick" and the "spacefill" rendering capacity of the current mspace window (for any structure)





Regards,


Gabor

Dear Gabor,

Quote:

It is not necessarily better. The Clean3D invoked on opening mview3D should produce a reasonable 3D structure

OK. And my idea that the two possible conformers must have nearly the same energy is confirmed by the Dreiding energy calculation: Both have 22.5 kcal/mol .

Quote:

MarvinSpace visualization even with the current limitations can come handy for medium sized structures.

OK. But it seems to me "Open Marvin Space" and "MarvinSpace itself" are different programs, at least different implementations. It makes unnecessary problems to explain in teaching that these two "Marvin Space" are different. And it produces unnecessary confusion.





So I suggest to give the MarvinSketch option "Open Marvin Space" a different name until it is implemented in the way you are planning to do. F.e. "MarvinMol3D" or so. This could be done in the next release, may be two or three JAVA code lines.





I attach my test molecule once more, opened with "MarvinSketch - View - Open MarvinSpace". It looks fine, just as in Jmol.





Regards, Hans-Ulrich

Hi Gabor,


to continue our discussion I have one example more, see the screenshot.





It is a deeply blue colorant and I have used the dearomatize method you showed in the post above.





But the nitrile groups look awful.





Triple bonds in MarvinSpace should be a developer task to be done very early.





It would be a Marvinspace feature other programs don't have.





Regards, Hans-Ulrich

Hi Hans-Ulrich,





we understand and admit the importance of proper visualisation of triple bonds, however, at present we do not have sufficient developer capacity to implement this feature. We may consider it in the longer term only.





Apologies.





Regards


Miklos

Hi Miklos,


I will show with another example, how important triple bonds can be. I cite http://en.wikipedia.org/wiki/Ethinylestradiol


"


Ethinylestradiol, also ethinyl estradiol (EE), is a derivative of estradiol. Ethinyl estradiol is an orally bio-active estrogen used in almost all modern formulations of combined oral contraceptive pills (the Pill). It is one of the most commonly used medications.





While estradiol is readily absorbed when taken orally, it is also quickly inactivated by the liver. Substitution at C17 of the estrane steroid with an ethinyl group proved to provide an estrogen that is much more resistant to degradation and paved the way for the development of oral contraceptives.


"


The ethinyl group is very important. It inhibits the oxidation of the hydroxy-group at C17 and is tiny enough not to trouble the ligand in the protein. So it is important to show the ethinyl-group.





MarvinView shows the triple bonds. So the algorithm to do it, is already in the Marvin software.





And the situation is not difficult. In organic chemistry, there are only two elements having triple bonds, C with two neighbor atoms and N with one neighbor atom. That's all.





And other programs, f.e. GChem3d, show triple bonds.





It' s a point of priority.





Best regards, Hans-Ulrich

Dear Hans-Ulrich,





the algorithm that is present in MView/MSketch cannot be applied or re-used at all in MarvinSpace for various reasons that I do not detail here. One important aspect is that MSpace is based on OpenGL while MView is not. If you are interested I can show you some code excerpts to demonstrate the huge difference.





It is not significant which type of atoms can form triple bonds in organic molecules, only the particular spatial arrangement of the three line segments or cylinders is what we need to work on.





I agree, and I have already admitted in my previous post, that the graphical notation of bond type can help user understand the chemical structure. Yet, the local geometry, that is, the relative spatial position of bonded atoms is the primary source of information: in this case the linear arrangement of atoms. Drawing bonds is not 'realistic' it's just nice. Being able to draw triple bonds is even better.


I would consider the situation critical if mspace was not able to draw atoms that have a triple bond. But that's not the case, so I cannot assign high priority to triple bond drawing - as I explained in my previous post, we have very limited resources allocated for mspace development at the moment. This, however, may change in the future in which case triple bond drawing will be implemented.





I created a task for that work, however, the activity has not yet been scheduled and has not been assigned to any developer. But it's in the pipeline, I can assure you. This also means, that you will receive a notification in this forum thread when the feature is implemented.





Thank you for your understanding,


sincerely,


Miklos

Dear Miklos,





OK, it's a difficult thing. I appreciate the high quality the MarvinSpace pictures show and now I know it's OpenGL !





But the pictures should also give information. And this information must be complete. As shown in the attached example (also important in the contraceptive pill chemistry), the information about bonds is only given for double bonds. I intentionally didn't use the "dearomatisation workaround" because the non-informed user cannot use it either.


So loading the structure you see only the "normal" double bonds und you get a wrong picture of the molecule !





The consequence should be: Before it is not possible to show all multiple bonds immediately after opening a molecule in MarvinSpace there should all multiple bonds shown as normal connection bonds, i.e. double bonds also as normal connection bonds.





It is better to have less information than an information which has defects.





So, the presentation of double bonds must be blocked in the code until there are also triple bonds and also an automatic presentation of aromatic rings.





Yes I am interested to look in the code. I think you have my email address.





Sincerely, Hans-Ulrich

Dear Miklos,





first I attach the display of gnome chemistry software gchem3d for the xyz-file for mifegyne.


The multiple bonds can immediately be seen, without any information in the file, only the cartesian xyz-coordinates of the atoms are given in the input.





I must repeat: As long MarvinSpace doesn't have the feature to show all multiple bonds, double and triple (and aromatic also as double bonds, without unknown workarounds) , it should not show only a part of them.





Someone doesn't know the stucture of a molecule given as xyz-file. So he uses MarvinSpace to see the structure and he sees in the mifegyne example as above, simply a wrong structure.





No information is better than information which is not complete: This leads to misunderstanding.





You offered me, to show me some code. I am very interested.





Sincerely, Hans-Ulrich

Hi,




thanks for these interesting details.


The code excerpts are not forgotten, but right now we cannot deal with this issue.




You'll be contacted later.




Miklos

Hi Hans-Ulrich,

Can you please check the triple bond visualisation in MSpace sometime? We are curious to learn your thoughts about it.

BR

Miklos

Hi Miklos,

having many students tutorials (using also Marvin) I have now time to answer you:

It's really good, to have now a good display of aromatic bonds and triple bonds in MarvinSpace !

So I have repeated the examples from the earlier posts here in this topic.

First the mifegyne example: Starting with the 2D-Formula shown in the attached image, followed by "Clean 3D with Hydrogenize" MarvinSpace (now called directly from inside MarvinSketch) shows a really good 3D representation of the molecule. Now the aromatic ring is shown and the essential triple bond, too.

And now there is another good result: The amino group is no more orthogonal as in earlier Marvin versions (see the image in the earlier post of this topic), now the amino group has the correct geometry and is in conjugation with the aromatic ring.

But, there are still problems with MarvinSpace. I will show these in the following post, to seperate it from this one.

Best Regards, Hans-Ulrich.

Hi Miklos,

now the problematic molecule from the earlier posts here: The Benzochinodimethan, see the attached image.

The nitrile group and the aromatic benzo group are now correctly shown. But now the chinodimethan system has an aromatic ring: wrong - it must have double bonds as a chinoic system. (The tool "Aromatize" was not used!)

To analyse that further I have constructed benzochinone itself, benzochinonmethide and benzochinodimethane: The second attached image shows, the chinone and the chinonmethide are shown correctly, only the chinodimethane shows an aromatic ring and double bounds going outside, producing five-valent carbons.

This remembers me to another topic about multiple bonds. I will continue my evaluation efforts.

Best regards, Hans-Ulrich

Dear Hand-Ulrich,

 

Thank you for the kind words and especially for the thorough tests and for the separation of problematic cases.

I checked the aromatisation of Benzochinodimethan in MSketch. The General method, which is used in MSpace, converts the chinoic system to aromatic, while the Basic method does not. So, in my understanding this is rather an aromatisation problem - however, I admit, MSpace should be able to switch between aromatisation methods similarly to MSketch, but at present this is not yet supported.

If you agree with me, that the problem of the aromaticity of the chinoic system is an aromatisation problem, then I will discuss this with the owner of that product and either get back to you with his response, or I put you two in touch.

Regards

Miklos

Hi Miklos,

Yes, I agree with you. It's obviously a problem of the conversion of double bonds to an aromatic form. As you see in the attached image, the two methods converting to aromatic forms give different results for the benzochinodimethane. The method "convert to general aromatic form" gives the wrong result. This remembers me to long discussions here in this forum about "Aromaticity" and I have still the question: What is the difference between "basic" and "general" aromatic forms ? It seems to me, the "general" aromatic form is something the chemoinformatic users of Marvin need to find conjugated systems being able to change to aromatic forms, and that's OK.

But MarvinSpace is not used in this way. It is used to show molecules in 3D and also to show the bond structures. And carbon having five bonds should be avoided.

Please change the default setting of MarvinSpace to convert to basic aromatic forms.

Best regards, Hans-Ulrich

Dear Hans-Ulrich,

I consulted with the developer of the aromatisation module and he admitted that the general aromatisation mode misinterprets this compound as you pointed out. This will be fixed.

That also implies, that after the fix marvinspace will properly display the structure, thus there's no immediate need to replace the default aromatisation mode in marvinspace.

Regarding aromaticity issues:

- you may find this forum area useful to discuss aromaticity problems

- you may wish to consult with the user's manual to learn about the differences between the basic and general aromatisation modes.

Regards,

Miklos

 

Hi Miklos,

after nearly two months I will reply to your last post here.

The topic you are mentioning has the title Aromaticity? - Call for opinions and in this topic you will find my 6 posts discussing the aromaticity problem.

At the end, Tue Mar 30, 2010, there was the consent to make the basic aromatization method as the default and the method one is choosing is saved.

Why is MarvinSpace using other methods and other defaults as MarvinSketch ?

Are there really different program methods used in "MSketch Clean-3D with Hydrogenize" and MSpace ?

And: Why does MSpace not save the settings of the user ?

There are more examples MSpace doesn't handle the 3D structure correctly, while Clean 3D can do it. May be, I will start a new topic about this.

Best regards, Hans-Ulrich

 

 

Hi Hans-Ulrich,

Regarding default aromatisation in MarvinSpace, in version 5.7 basic will be the default one.

Regarding saving settings... that's a longer term work, not yet scheduled.

Looking forward to learning your comprehensive analysis about the 3D structure handling problems!

 

Kind regards,

Miklos