MSpace bugs: triple bonds, sp carbon, phenylgroups

Changed Sep 27, 2010: The following posts document several bugs of MarvinSpace called from inside MarvinSketch (applet and desktop).

Hi, some comments here:

In two old topics (ftopic4210.html Oct 16, 2008 and ftopic5884.html Feb 22, 2010) there were the requests, MarvinSpace should be able to show triple bonds. Having now 5.3.8 this is still not implemented, see the attached screenshot showing the 2-butyne mentioned above.

And I cannot see the problem. The code to program this must be present, MarvinView is able to show triple bonds. So the code can be transferred from MarvinView to MarvinSpace. And it should be done to both versions, the "mini"-version included in MarvinSketch and the full version in MarvinSpace itself.

Regards, Hans-Ulrich

Dear Hans-Ulrich,

Your post was moved to a new topic since it adresses a different issue.

The code to program this must be present,

MarvinView is able to show triple bonds. So the code can be transferred

from MarvinView to MarvinSpace.

I would like to note that MarvinSpace is a different software, above mentioned "code transfer" is not a possible way to implement the suggested display functionality. This question is detailed in https://www.chemaxon.com/forum/ftopic4210.html

Regards,

Gabor

Hi,

the attached screenshots are produced in the following way: The molecules are drawn in MarvinSketch, without any "Clean" or other modifications.

Then the molecules are opened with the in MarvinSketch included tool "View - Open MarvinSpace". The result can be seen in the upper part of the screenshot.

Then "View - Open MarvinSpace" is opened once more and here "Right mouse click - Draw Type - Hydrogens" is activated. This result is shown in the lower part of the screenshot.

Now to the two C3H4 isomers :

1) In the upper MSpace window propyne has no triple bonds. One cannot see the sequence of the atoms.

2) In the lower screen the methyl group of propyne is displayed, but the left hydrogen of the yne-group has a wrong structure. It should in the inline with the CC-bonds.

3) In the lower screen all hydrogen of the allene are in one plane. That's wrong. These CH2-groups should have "orthogonal" hydrogens.

More in the next post.

For these C2H3N isomers "View MarvinSpace" shows the following:

The display does not correspond to the display in MarvinSketch. Having two lines cannot be avoided (But this is not a series problem and is only explained here).

1) Once more CC- ore CN-triple bonds are not shown. One cannot identify which molecule is acetonitril and which is amino-acetylene. One cannot see nitrile groups.

2) In the lower part, the hydrogen at the aceytlen part of amino-acetylen should be inline with CC.

The ketenimine has the correct orthogonal structure for the hydrogens.

More in the next post.

The last molecule of this report is carbondioxide CO2 (A rather important molecule).

Here even the douple bonds are not shown by MarvinSpace. Obviously is MarvinSpace here not able to handle the sp-hybridized carbon in the middle.

Let's stop here. I hope these severe bugs will be fixed soon.

Regards, Hans-Ulrich

Hi,

testing "MarvinSketch - View - Open MarvinSpace" with phenylisocyanate and cyclohexylisocyanate the result is shown in the screenshot.

1) The isocyanate group is correct with two double bonds. Here the sp-carbon in the middle is obviously handled correctly.

2) But the phenylgroup is shown without double bonds. I have already mentioned this defect of MarvinSpace about two years ago in the topic https://www.chemaxon.com/forum/ftopic4210.html

3) Adding "Draw Type - Implicit Hydrogens" inside MarvinSpace shows, the two six membered rings are analysed correctly, only the double bonds in the phenylgroup are lacking. And here the "workaround" described in the topic 4210 mentioned before is not applicable.

When will the severe bugs documented in this topic be fixed ?

Regards, Hans-Ulrich

Hi,

thank you for the new bug reports and the valuable test cases. Both triple bond and aromatic ring visualisation tasks are in progress. They are planned to be released in 5.4.1 towards the end of this year.

You'll be informed in this forum topic and early developer version will be made available for evaluation for you: we rely on your valuable feedback, we are open to learn your thoughts on 3d structure visualisation. Stay tuned.

BR

Miklos

Hi Miklos, thank you for your positive answer.

Continuing my contributions I analysed some unsaturated sulfur compounds, see attached screenshots (Carbon disulfide was a very important building block in my experimental research many years ago ...). The situation is obviously the same as for the corresponding oxygen compounds posted here earlier. And there is the question, why is MSpace able to show the double bonds in the nitrogen compounds and is not able to show these in the oxygen and sulfur compounds.

All these compounds have a sp carbon in the middle, having two double bonds to both sides. I think it's important for programming to do the following: First to count the number of neighbors each atom has and then to determine the number of bonds by using the valencies of the atoms and the valencies of the neighbors. There are to many compound classes to classify the bond orders.

Best regards, Hans-Ulrich

Thank you Hans-Ulrich,

useful as always. We will consider these examples among many others.

Miklos

Hi Miklos,

evaluating the new Marvin 5.5.0.1 according to the theme of this topic my first impression was: triple bonds are now visible in MarvinSpace, but what's about allenic bonds? For the sp-carbon in allenic type bond systems there are still no double bonds.

But, after moving the molecule in 3D I found it. In the original molecule orienation, you see the two bond sticks from the side, so the second is hidden behind the first. The attached example handles the important molecule carbon dioxide CO2. The upper MarvinSpace view shows how MarvinSpace starts to show the molecule. After rotating the molecule around the horizontal x-axis about 90 degrees, the double bonds are visible !

The molecule viewer Jmol shows double bonds always, independant how the molecule is rotated.

Best regards, Hans-Ulrich

Hi Hans-Ulrich,

Interesting indeed! So the way how double bonds are drawn is fine, I mean that they are co-planar, but the orientation (view) of the structure should be changed. Is this what you mean? So you are not suggesting that one of the double bonds should be turned 90 degrees to make sure that in all view aspects at least one of the is visible....

BR

Miklos

Dear Miklos,

I agree, a double bond has a direction corresponding to the planar sp2-carbon. But which direction? To the direction of the pi-orbitals, orthogonal to the sp2-plane ?

I think the best way is to do it like Jmol. The two bond sticks of a double are always orthogonal to the viewing direction, so these are always visible, independent of the molecule plane. It's more important to see the double bonds at all, as to get a wrong information about the bonds, just as in the carbon dioxide above.

The double bonds and the triple bonds should be always shown parallel to the image plane.

And I think it should be easy to program this. In the present version the two sticks are parallel to the molecule plane which rotates with the corresponding molecule. To have a fix plane is easier to program.

And the distance between the sticks should be a little bit larger. So these are better visible in large molecules.

And BTW: In carbon dioxide the two double bonds itself are orthogonal. The carbon in the middle is sp-hybridized having two orthogonal pi-orbitals.

Best regards, Hans-Ulrich

Thank you, and learning a lot~~~

Hi Hans-Ulrich,

The double bonds and the triple bonds should be always shown parallel to the image plane.

Hi Miklos,

discussing the "double bond problem of MarvinSpace" with colleagues we all found: Showing the two double bond sticks in the sp2-plane is not good. And we found several model kits to construct an ethylene molecule, please see the attached image.

Normally these kits use bent bonds to build the correct geometry (In a molecular orbital model one can interpret these bent bonds as linear plus and minus combinations of the sigma orbital and the pi orbital).

But showing the two sticks orthogonal to the sp2-plane has the same problem as showing in plane. Looking at the plane you don't see the two sticks being behind each other.

So we all found the Jmol solution is the best way to show in a 3D display where a molecule has double bonds and triple bonds.

Kind regards, Hans-Ulrich

PS: Of course, the model kits have trigonal atoms, too. But these give the option to show pi-orbitals, and these are orthogonal to the sp2-plane, too.

Hi Miklos,

to emphasize the Marvinspace problems displaying double bonds I will show here one example more: Mycomycin OC(=O)C\C=C\C=C/C=C=CC#CC#C a naturally occurring antibiotic isolated from the bacterium Norcardia acidophilus. The attached mol file includes the 3D geometry produced using a quantum chemical program with the semiempirical AM1 method.

The upper attached image shows the molecule in 3D exactly in the 3D orientation as calculated and documented in the mol file. In the MarvinSpace display the triple bonds are shown but not very impressive, and the double bonds are hidden. These are displayed along the sp2-plane as discussed in the posts before. Of course you can rotate the molecule in 3D but only in special positions you see both double bonds and triple bonds.

By chance I found today a very practical Marvin tool: convert2image

https://www.chemaxon.com/forum/ftopic1622.html 

This program produces an image showing the molecule exactly in the 3D orientation defined in the mol file. And it shows all double and triple bonds orthogonal to the viewing direction !

And MarvinView 3D shows all bonds in all 3D orientations, too.

MarvinSpace should show all bonds in the same way.

Best regards, Hans-Ulrich

Hi Hans-Ulrich,

thank you for the nice example, I understand your requirement. MarvinSpace will support this kind of display option in the future.

Best regards,

Miklos

Hi Miklos,

having now Marvin 5.6.0.0 I immediately tested MarvinSketch - View MarvinSpace. Pasting the SMILES string from my last post here produced in MarvinSketch the exact structure of the highly unsaturated mycomycin. Super!

But the instantanously produced 3D display in MarvinSpace is as inaccetable as the examples shown here earlier.

To have a correct and acceptable display is not an option, it will be a bug fix.

This topic here is now two years old coming from Marvin 2.3.8 and the title does no longer correspond to the actual discussion. But before opening a new topic about double bonds I will discuss the problem of hydrogens.

Regards, Hans-Ulrich

Hi Hans-Ulrich,

A new bond rendering was implemented which enables the bond rank visualisation according to your flavour. This will be released in 5.8 (or perhaps sooner in 5.7.?).

Are you interested in downloading a nightly build to give it a try?

BR

Miklos

Hi Miklos,

that's good to read ! Oh, yes. Let me download a trial version. So I will going to have no more problems in our student courses. And I can test it !

If you want to do it directly: I think you have my email address.

And: I need to have the Linux version !

Thank you and best regards, Hans-Ulrich

Hi Miklos,

having now Marvin 5.7.0 I must post one ugly example more: The planar 3,4-dimethylidenecyclobut-1-ene, see attached screenshot.

We are also using Marvin for first semester students. And these presentations are really problematic.

Best regards, Hans-Ulrich

Dear Hans-Ulrich,

This feature was not released in 5.7 but will be definitely part of 5.8. See attached image.

Thank you for your understanding.

Miklos

Hi Miklos,

looks very fine ! And what's about triple bonds ?

In my post Sat Sep 10, 2011 10:15 in this topic you may find the mol-file of mycomycin. And you may test the new version with this molecule having nearly all possible multiple bonds.

And one question: May I have a Linux pre-release. Conjugated pi-systems are an important topic of our actual student courses (dealing with colorants).

My mailbox is large enough ...

Best regards, Hans-Ulrich

Hi Hans-Ulrich,

I attach the picture of mycomycin. Regarding the trial download I sent you an email.

Miklos

Hi Miklos,

super ! We are thrilled.

Thank you.

Best regards, Hans-Ulrich

Hi Miklos,

something is wrong. I downloaded today the trial version you mentioned in your email, but MarvinSpace still shows the same image with bad double bonds as in my earlier post Sat Sep 10, 2011

Regards, Hans-Ulrich

Hi Hans-Ulrich,

please check Display->Options->Visualizers->Front facing bonds

This is not the default behaviour.

(I know your next request: enabling the saving op options! ;-) )

BR

Miklos

Hi Miklos,

thank you for the hint, but my next request goes deaper: The new and much better display should be the default. Is there anyone who wouldn't like to see all multiple bonds ?

Attached is one example from earlier posts here. But we will test all the examples here , too.

Best regards, Hans-Ulrich

my next request goes deaper: The new and much better display should be the defaul

Indeed, others believe that the view (the 2D projection of the 3D structure) should not change when the 3D structure (which is believed to be structurally rigid) is tumbled (i.e. looked at from different aspect).

Regards,

Miklos

Hi Miklos,

having now Marvin 5.10.0 the display of double bonds is still inaccetable. The attached image shows a benzene molecule coming for a quantumchemical calculation. The result was read from a xyz-file, the double bonds have been added in MarvinSketch and the display in MarvinView is perfect.

But what's about the display in MarvinSpace ? Here the third double bond at the bottom is invisible.

And in this context I still don't understand: "Indeed,
others believe that the view (the 2D projection of the 3D structure)
should not change when the 3D structure (which is believed to be
structurally rigid) is tumbled (i.e. looked at from different aspect)." Benzene is planar and all pi-bonds have the same direction.

Yes, the option "Front facing bonds" is off. You have to remember it,
and you have to show the students, how to switch on this
option in MarvinSpace.

And every time you start MarvinSpace once more, you have to switch it on. Even when coming a second time from MarvinSketch. MarvinSpace still doesn't remember your settings (There are more settings, for which it would be comfortable to have the settings stored).

Let me repeat: Front facing bonds should be the default. Is there anyone who wouldn't like to see all multiple bonds ?

Best regards, Hans-Ulrich

Dear Huns-Ulrich,

This invisible bond (and actually, that fact that each double bond is in a different plane) is clearly a bug. We will fix that.

Regarding the default option I'm curious to learn the result of your poll (that you either conduct here in our forum, or elsewhere). In version 5.11 we will adjust the default accordingly.

Kind regards

Miklos

Hi Miklos,

before answering the question about the "poll" one example more. Coming from an quantumchemical calculation we tried to produce an impressive image of the fungicide isotianil. MarvinView3D shows a good presentation of the multiple bonds, but in the MarvinSpace presentation some multiple bonds are invisible, and the option to show visible multiple bonds is not stored and has to be set once more ...

About the "poll" the answer is: Since Jan 2012 more than 900 visitors read the request:

"The new and much better display should be the default. "

And there was no opposite answer.

The algorithm is installed and MarvinView3D does it perfectly as long as I know it.

So: Please change the default in the next MarvinSpace version. MarvinSpace has some very useful options which are singular. And so it would be fine to fix the bugs.

Best regards, Hans-Ulrich

Hi Hans-Ulrich,

Thank you for the example structure. We will meet your requirement in the next release of Marvin. However, at the moment I cannot yet tell which solution will be supported (default behaviour changed or view options stored and re-used). We will discuss this internally and then I get back to you.

Regards,

Miklos

Hi Miklos,

please install the visability of multiple bonds as default !

Let's look at the following situation: In our training courses we have to instruct each semester hundred and more students to set in MarvinSpace "Display-Options-Visualizers-Front facing bonds" to [v]. Some days later they use another environment, at home or the notebook, and here they must do the same. This will not work.

And once more the question: Is there anyone who wouldn't like to see all multiple bonds ?

Best regards, Hans-Ulrich

Dear Hans-Ulrich,

Please be assured that it will be the default behaviour in release 5.11 and after.

Best regards,

Miklos

Dear Miklos,

super, now we have it !

Marvin 5.11.0 beta 23 now shows all multiple bonds. And it is the default option !

I've attached some examples discussed here in this topic. Two simple molecules and one molecule having all essential bond types: mycomycin. All double bonds, triple bonds and especially the allene bonds are now visible.

Now MarvinSpace is fit for student courses.

Thank you very much.

Best regards, Hans-Ulrich

Dear Hans-Ulrich,

Thank you for the voluntarey beta-testing and for the positive feedback. Please keep pushing us in the right direction in the future.

I'm glad to hear that now you can and you will use MarvinSpace in your chemistry classes.

Best regards,

Miklos