MSpace bugs: Wrong hydrogen geometries and bond displays

Hi,

MarvinSketch has now the tool to show instantaneously in View-MarvinSpace the molecule one is drawing: A really excellent tool. And it works normally very fine with larger molecules such as sugar, as shown in my last post.

But calculating the geometry for hydrogens in small molecules View-MarvinSpace shows severe failures. The attached image shows methane as the smallest molecule in organic chemistry: The hydrogens are in a planar structure instead of the well known tetrahedral 3D structure.

But after using the "Clean 3D Fine with Hydrogenize" structure tool, View-MarvinSpace shows the correct tetrahedral structure. Obviously there are severe failures in the 3D calculation algorithms of MarvinSpace.

Of course, the structure of methane is well known for chemists. But this failures contradict to use MarvinSketch and View-MarvinSpace for newcomer chemistry students. I showed this problem an assistent of my computer chemistry courses: She was shocked to see the methane structure and the structures of ammonia and water, shown in the next posts here.

Retracing these bugs I found it was also in the oldest version I have still installed and running here: 4.1.13.

Why does View-MarvinSpace not use the same 3D-algorithm as Clean 3D ?

Regards, Hans-Ulrich

Continuing the structure problems of MSpace here comes the problem with ammonia NH3. Of course it is not planar but has a pyramidal structure, as produced by Clean 3D Fine with Hydrogenize.

And now water H2O. It is not linear, but has a bond angle of about 104 degrees. Once more Clean 3D calculates the correct structure.

Hydrogen sulfide H2S has in View-MarvinSpace a wrong linear structure, too. But I will not attach the image here.

There are more MSpace problems with hydrogens in small and larger molecules, but for today I will stop here.

Regards, Hans-Ulrich

The attached image shows, MarvinSpace produces ethane in the eclipsed conformation which is a transition state and not a minimum. Once more, Clean 3D with Hydrogenize is able to calculate the correct structure and now MarvinSpace is able to display it.

The second image shows n-hexane. Now the "all-staggered" conformation is calculated correctly by MarvinSpace, too. Only the end methyl groups doesn't have the correct conformation. These are between eclipsed and staggered.

MarvinSpace has problems to display double bonds. I have already shown these in an other topic

https://www.chemaxon.com/forum/ftopic6633.html

In this topic the double bond display for mycomycin has an other direction as for ethylene, here. For ethylene it is now orthogonal to the plane of this molecule.

In cyclopentadiene the double bond display is different left or right side and the display is now a little bit in the plane of the five membered ring.

Vinylcyclohexadiene has a vinyl group which is not planar (Clean 3D with Hydrogenize produces here a correct structure, too).

The display of double bonds should be organized in such a way so that you can always see it. Please look in the topic mentioned before.

And now acetylene. In MarvinSpace it is not linear and from some directions you see the triple bond as double bond.

The structure MarvinSpace is displaying for hydrogen cyanide HCN is not linear, too.

May be these failures are not important for chemoinformatics. But using Marvin for education is very problematic. Many people are believing all results a computer is showing them.

Please fix these bugs as fast as possible.

Best regards, Hans-Ulrich

Hi Hans-Ulrich,

Please test the alternative visualisation of double and triple bonds in the 5.7beta release.

Regarding the wrong geometry of the acetylene, this bug will be fixed in a forthcoming release, but not in 5.7 I'm afraid. Apologies for the inconvenience this bug caused (and still causes).

Best regards,

Miklos