Problems in geometry tool

Hi Marvin developers,

further evaluating MarvinSketch in context of geometry calculations I found in the geometry tool some irritating displays. In the attached screenshot for the simple linaer molecule acetylene there are two "area"-circles. My interpretation is: The yellow circle corresponds to the "Maximal projection area". But, for the green circle, corresponding to the "Minimal projection area" I have some questions:

1) Why is the green circle not orthogonal to the axis of the linear molecule ? I think the "Minimal projection area" should be in this direction.

2) Why is the center of the green circle not at the center of the molecule ?

Searching for help for this geometry tool I didn't find any color descriptions neither in MarvinSketch "Help" nor in the online "User's Guide". The correspondance of "yellow" and "green" should be given directly in the "Geometry" window.

There are more problems in this tool, but I will stop here and wait for an answer.

Regards, Hans-Ulrich

Hi Hans-Ulrich,

thank you for the deep evaluation of the projected surface area calculation. Please find comments inline.

Regards,

Adrian

HUWagner wrote:

Hi Marvin developers, further evaluating MarvinSketch in context of geometry calculations I found in the geometry tool some irritating displays. In the attached screenshot for the simple linaer molecule acetylene there are two "area"-circles. My interpretation is: The yellow circle corresponds to the "Maximal projection area". But, for the green circle, corresponding to the "Minimal projection area" This is the correct interpretation. I have some questions: 1) Why is the green circle not orthogonal to the axis of the linear molecule ? I think the "Minimal projection area" should be in this direction. The underlying method is quite complex and uses 2D gaussian discs for representing the projected area of a 3D molecule. This algorithm uses some neglections in order to produce reasonable runtimes for drug size molecules. These neglections however produce such behaviour for linear "rod-like" molecules but have minor effects on the area of the molecules having various other geometrical features. We have an other algorithm in mind which solves this issue but unfortunately it will not be scheduled in short term. 2) Why is the center of the green circle not at the center of the molecule ? This may be a bug, I will check and correct it. Thank you. Searching for help for this geometry tool I didn't find any color descriptions neither in MarvinSketch "Help" nor in the online "User's Guide". The correspondance of "yellow" and "green" should be given directly in the"Geometry" window. You are absolutely right we will add the legend for the color coding in the help soon. There are more problems in this tool, but I will stop here and wait for an answer. Regards, Hans-Ulrich

Hi Adrian, one very good quality of Marvin is the very active discussion !

Before continuing the problems one question more: Are my correlations of the green and yellow arrows and circles shown in my screenshot correct ?

And: For the benzene example the green circle is once more not in the middle.

Best regards, Hans-Ulrich

HUWagner wrote:

Hi Adrian, one very good quality of Marvin is the very active discussion ! Before continuing the problems one question more: Are my correlations of the green and yellow arrows and circles shown in my screenshot correct ? And: For the benzene example the green circle is once more not in the middle. Best regards, Hans-Ulrich

Hi Hans-Ulrich,

your correlations are absolutely correct. Thank you for the circle position bug. Very likely it is the same situation as before.

Best regards,

Adrian

Hi Adrian,

in the attached picture there are some analyses for the van der Waals surfaces of actylene. The two projections (procuced using MarvinSpace) correspond to the symmetry of the molecule. Here it's easy to calculate the two significant "Length" values.

"Length perpenticular to min area" (green) = 5.44 A is the sum of the molecule length (given by MarvinSpace) plus 2 times the van der Waals radius for hydrogen (1.20 A). This value is higher as in "Marvin Geometry" because the geometry is incorrect in this calculation, see the picture in my first post of this topic.

"Length perpenticular to max area" (yellow) = 3.40 A is 2 times the van der Waals radius for carbon (1.70 A). This value is the same as in "Marvin Geometry" having the same correct geometry.

I have changed the symbolism for these lenghts, because a "length" is not an arrow and has no direction. I suggest to use a line with two marks at the end, as shown in the attached picture.

In a later post here I will also have a suggestion where to locate the "length" in the molecule.

But before doing this I will analyse the "Geometry" calculations for some "more complicated" molecules.

Regards, Hans-Ulrich

Hi Adrian, now the planar molecule formaldehyde having C2v symmetry.

In the first upper picture there is the original MarvinSketch Geometry calculation for formaldehyde CH2O. The molecule is exactly positioned in the drawing plane: The yellow arrow is not visible. Two results are obviously wrong:

1) As before the green circle for the "minimal projection area" is not orthogonal to the axis of the molecule. The same bug as for acetylene.

2) The molecule is not in the middle of the "maximal projection area".

The second pictures shows some hand drawing of the van der Waals spheres for the 4 atoms. These are of course overlapping, but the yellow (orange) circle for the "maximal projection" area does not include the whole area. The position of this circle is wrong.

My next post will analyse a real 3D molecule.

Regards, Hans-Ulrich

Hi Hans-Ulrich,

thank you for the enthusiasm and the detailed test of formaldehyde. The translation of the center of the yellow circle roots back to the same bug as before.

However the tilt of the green axis from the obvious direction is the result of the neglections which are currently used and appearently needs to be revisited. 

Best Regards,

Adrian

Hi Adrian, now a 3D molecule, a little bit more complex: Androstan.

Here the orientation of the area circle is better. The large molecule is surrounded.

Some suggestions for the 3D-presentation of the results:

1) Different colors for the "area circles" and the "lengths". These results are not directly correlated as interpreted using only yellow and green.

2) The colors should be assigned in the little table of the results, f.e. as shown.

3) The "length" line should show the length inside the molecule. The until now used arrow represents one direction outside.

If you want, you may use the second picture for new guide presentations.

Regards, Hans-Ulrich

PS: The pictures are produced using GIMP. A marvelous program.

Hi Hans-Urlich, 

thank you for your valuable suggestions. #3 is something that I have already thought of but #1. and #2. are something new and really reasonable to implement. Once this task is scheduled again we will consider the implementation.

Kind regards 

Adrian