Transfer geometry from Plug-in to Sketch

User b39cf20463

23-06-2010 11:52:17

This relates to the Geometry and Conformation plug-ins, but I felt the issue was more general. Marvin version 5.3.1


I am using the sketch applet within a web page, then using the geometry optimiser on a newly drawn structure. Then I use javascript to extract the structure from the applet for uploading to the server.


The problem is that the geometry optimiser spawns a new Java window that I do not know how to access from Javascript. The conformation plug-in has a "select" button that returns the chosen conformer to Sketch, so it is possible to do what I want but I cannot find any corresponding option within the geometry plug-in output window.


At the moment, I can take the optimised structure back to Sketch by cutting and pasting it, but I do not want my users to have to do this if at all possible.


Can you suggest the best way for me to do this?


Many thanks.

ChemAxon e08c317633

24-06-2010 17:16:21

Please see the following code example: MarvinSketch Example - Visualizing 3D structures
This example shows how to call clean3d to convert the input structure to 3D.


Does this solve your problem?


Zsolt

User b39cf20463

25-06-2010 09:37:40

Thank you for this suggestion.


In many respects, Clean3D does do what is required, but I have concerns about the cleaning algorithm. I don't know what it is doing under the hood, and how far it can be trusted with interesting systems.


The structures will be fed into quantum codes that are very sensitive to initial geometry, so I was looking for the most rigorous structure minimisation possible through Marvin. I will take it as read that such a geometry can only be returned by copy-paste then (or switching to the Bean distribution and using the API).

ChemAxon 1b9e90b2e7

06-07-2010 06:55:08

Hi Kieron,


Clean3D uses an in house developed fragment based building process:


The input 2D structure is being cut to small fragments: the initial fragment size is one or two atoms.


These fragments are optimized with ChemAxonForceField(CAFF). CAFF originates from Dreiding force field, which is used by for example in Gaussian to obtain meaningful input structure for quantum chemistry calculations. The original Dreiding was enhanced at least in two respects:
1.  Any kind of molecules with any atomtypes can be used
2. The potentials for common chemical moieties was enhanced a lot by fitting it to quantum chemical results.


So when we have small optimized 3D fragments we fuse them to get bigger ones. Multiple conformations are stored to one fragment so if a steric clashes or proximity errors appear a new fragment conformer is used. So the quality of the 3D structure is analized during the building
process. Only optimized structure is returned.


Clean3D returns acceptable conformers for more than 99% of our test cases. (NCI, zinc druglike...)


I encourage you to use Clean3D  structures for quantum chemical input and really looking forward to your experience and possible problematic structures.


Hint:


The default operation usually returns acceptable structures however if you wish to go for more quality you may try:  Tools\Conformation\Conformers\ 
Hyperfine selected,
Lowest energy unselected,
Max number of confs ~10 or more.
Display Conformers selected.


 


Good luck,


Adrian