[Jmol-developers] [ jmol-Feature Requests-1032448 ] Calculation and display of various surface types

[SourceForge.net <no...@so...>]

Feature Requests item #1032448, was opened at 2004-09-22 04:00
Message generated for change (Comment added) made by hansonr
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=379136&aid=1032448&group_id=23629

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
Status: Open
Priority: 5
Submitted By: Christoph Steinbeck (steinbeck)
>Assigned to: Bob Hanson (hansonr)
Summary: Calculation and display of various surface types

Initial Comment:
In order to establish Jmol as a Chime replacement, the
calculation and display of various surface types would
be needed. 
The attached graphics have been take from teaching
module of the german project "Networked chemistry
studies" which is currently based on Chime but where at
least some people would love to get rid of it.

The file names "Chime Gasteiger charges" shows a model
calculated by the following chime command:


script="surface molsurface white 0.45; list molsurface
transparent; set 
mep distance 99.0; set charge function gasteiger;
calculate charges 
refresh; list molsurface color potential"

The other one named "Gaussian cubs" has been created
from a so-called cubs file produces by the
ab-inito-program Gaussian. The cubs file is also attached.

----------------------------------------------------------------------

>Comment By: Bob Hanson (hansonr)
Date: 2006-07-04 18:04

Message:
Logged In: YES 
user_id=1082841

Recent advances in Jmol functionality may make this less of
an issue. These comments refer to Jmol 10.x.19, in
http://www.stolaf.edu/people/hansonr/jmol/test/proto and
demonstrated and discussed at
http://www.stolaf.edu/people/hansonr/jmol/test/proto/isosurface.htm
and
http://www.stolaf.edu/people/hansonr/jmol/test/proto/isosurface.txt


1) You can now render proper solvent-excluded surfaces
(molecular surfaces) using the SOLVENT keyword.

  select (atoms-to-include)
  isosurface solvent <radius>

2) You can now render proper solvent-accessible surfaces
using the SASURFACE keyword.

  select (atoms-to-include)
  isosurface sasurface <radius>

3) You can set the resolution in grid points per Angstrom
using the RESOLUTION keyword. For a large system, 0.5 might 
be appropriate.

  isosurface resolution 0.5 sasurface 1.2

4) You can, of course, render these as dots, mesh, or solid
(fill/nofill) and either translucent, or opaque using those
keywords.

  isosurface s1 resolution 0.5 sasurface 1.2 mesh translucent

and you can change these properties later using another
isosurface or color command

  isosurface s1 opaque dots nomesh
  color $s1 translucent red

5) You can capture a JVXL equivalent (see
http://www.stolaf.edu/people/hansonr/jmol/docs/JVXL-format.pdf)
in order to deliver this surface over the web in a very
compact manner that renders quickly. (Using "show
isosurface" or just by looking in the Java console.) 

  isosurface "savedfile.jvxl"

6) You can map these isosurfaces (either the original or the
JVXL equivalent) using CUBE data.

  isosurface solvent 1.2 color "mydata.cube"

7) You can set the mapping range using the ABSOLUTE keyword.

  isosurface solvent 1.2 color absolute 0.1 1.0 "mydata.cube"

You can save this color-mapped object as a JVXL file as
well, so you only need to use the CUBE file once.

8) You can apply a variety of color schemes using the
COLORSCHEME keyword.

  isosurface colorscheme "roygb" solvent color "mydata.cub"

Other schemes include "rwb" (red-white-blue), "low"
(red-green), and "high" (green-blue).

9) You can reverse the color using the REVERSECOLOR keyword:

  isosurface reversecolor solvent color "mydata.cub"

So, in a nutshell, we've come a long way since this thread
began. 

Still to do: fake esp surfaces along the lines of Chime,
real esp surfaces using Jmol's newly built in molecular
orbital quantum mechanical calculation capability using ab
initio basis sets. What else? Advice is requested as to
priority and direction at this point. 

Bob Hanson


----------------------------------------------------------------------

Comment By: dcaffrey (dcaffrey)
Date: 2006-07-04 01:25

Message:
Logged In: YES 
user_id=1040751

Sorry for the delay in responding. 

Yes, the naccess links and references you provide are
useful. It may also be useful to look at the original paper
describing the method. Lee and Richards J Mol Biol 1971
55:379-400. It provides useful details on "sectioning" the
structure and computing the area for each atom slice
(equations 1,2,3)

Q1 Hydrogens are usually ignored.

Q2 The zslice == 0.05 but I'm not aware of any shortcuts for
this aproach.

Q3 The Lee and Richards approximation is generally accepted
and still commonly cited. While there are more sophisticated
approaches it has prevailed for some reason or another.

Q4 Agreed. The network science link is correct.

----------------------------------------------------------------------

Comment By: Bob Hanson (hansonr)
Date: 2006-06-15 10:05

Message:
Logged In: YES 
user_id=1082841

>From http://bioweb.pasteur.fr/docs/doc-gensoft/naccess/README
I see what you are talking about, namely:

1) An atomic accessibility file (.asa file) containing the
calculated
accessible surface for each atom in a PDB file, as well as
the assigned van der
Waal radii.

2) A residue accessibility (.rsa) file containing summed
atomic accessible
surface areas over each protein or nucleic acid residue, as
well as the
relative accessibility of each residue calculated as the
%accessiblity compared
to the accessibility of that residue type in an extended
ALA-x-ALA
tripeptide (for amino acids). See Hubbard, Campbell &
Thornton (1991) 
J.Mol.Biol.220,507-530. You can prevent this file from being
calculated 
if you don't need it.

Several questions come to mind:

Q: How is this done in light of the fact that the H atoms
are not generally present for biomolecules?

Q: The surfaces I'm setting up probably are too expensive
computationally for proteins. What shortcuts do people use?

Q: I took a relatively close look at the cdk code. It's
basically the first part of what we implement in the
Dots.java routine, where we use geodesic tesselation as done
here. It is only calculating the "single-atom-contact"
surfaces, so it would not be appropriate for actually
visualizing the surface, but I can see how it gives a nice
(very rough) calculation of the solvent-accessible surface
area and atomic "volume". Since we already do all these
calculations as part of dots rendering, I don't see why we
couldn't provide this sort of analysis as well. To see the
"dots" associated with this tesselation calculation, just 

http://www.stolaf.edu/people/hansonr/jmol/test/proto/dots.htm

load 1crn.pdb
set solvent 1.2
dots on
color dotsconcave black
color dotssaddle black

As you can see, there are a lot missing! I guess I'm quite
surprised that people settle for such a gross approximation.
But I guess if all you are doing is getting a number and not
actually drawing the surface, you might be satisfied with
this. Is this all people want in terms of calculation?

Q: Something I'm VERY surprised to see here is that the
"surface" for which the area is being determined is not the
one I've associated with the term "solvent-accessible"
surface. At least not the one I've ever seen rendered. I do
hope we are doing this right in Jmol! The area and volume
calculated are for the surface defined by the travels of the
CENTER of the solvent probe. Is that really the surface
people are talking about when they refer to
"solvent-accessible surface"? Not the CONTACT surface? 

oy. I see:

http://www.netsci.org/Science/Compchem/feature14e.html

yup, that's what it is. OK, I see. We are depicting not the
"solvent-accessible" surface but rather the "molecular
surface". 

The true "solvent-accessible surface" is (almost) trivial
compared to the molecular surface. 

I think we need to discuss this on the jmol-developer list.

Bob Hanson






----------------------------------------------------------------------

Comment By: dcaffrey (dcaffrey)
Date: 2006-05-18 22:21

Message:
Logged In: YES 
user_id=1040751

It would be nice if user could retrieve numbers from the
console  that are similar to the .rsa and .asa files
produced by naccess.  

Questions:

1) Do you intend to use the cdk implementation?
http://cdk.cvs.sourceforge.net/cdk/cdk/src/org/openscience/cdk/geometry/surface/NumericalSurface.java?revision=1.13&view=markup

2) Any updates on when this might be available?

----------------------------------------------------------------------

Comment By: Miguel (migueljmol)
Date: 2004-09-22 08:22

Message:
Logged In: YES 
user_id=1050060

Adding surfaces is certainly a priority. 

However, support for surfaces will come after the official
v10 release


----------------------------------------------------------------------

Comment By: Bob Hanson (hansonr)
Date: 2004-09-22 06:05

Message:
Logged In: YES 
user_id=1082841

I would like to see this a priority. Descriptions of the
functions can be found at

http://bip.weizmann.ac.il/course/prog2/graphics/chimescript2.html#set_mep_mlp_function

At the very least we should have something like:

surface molsurface white 0.1;
set mep distance 100.0;
set charge function gasteiger; 
calculate charges refresh;
list molsurface color potential red orange yellow green blue
darkblue;
list molsurface transparent;

"mep" here stands for "molecular Electrostatic Potential"

These are very useful. But I have found that Chime does a
poor job of rendering the colors. We could do much better.


Bob Hanson

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=379136&aid=1032448&group_id=23629