VMD-L Mailing List

From: Jan Saam (saam_at_charite.de)
Date: Thu Mar 23 2006 - 16:11:55 CST

As a first step of improvement wouldn't it be possible to implement the
following strategy in "measure fit":

Try algorithm 1 first and compute the RMSD. The RMS fit may fail but the
RMSD can't be cheated! If the RMSD after the fit is worse that some
default (which could be scaled by the atom number and adjusted by the
user though an option to the measure fit command) then try algorithm 2.
Now simply use the better result.

If it's true that the two algorithms fall down in different cases, we
have a good chance to catch a reasonable number of these failures. And I
guess it would only require a few lines of code.

What do you think?

Jan

John Stone wrote:

>Jan,
> VMD and PyMol are (or at least were) using the same exact
>RMS fit code. VMD actually has two implementations, both derived
>from Kabsch, so they fall down in about the same way, but in different
>cases. I agree it would be interesting to collect info from other
>people's experiences with other packages, but thought I'd mention
>that the PyMol / VMD RMS fit code should actually be the same.
>You can switch which code VMD uses by setting and environment variable
>that I don't recall off the top of my head BTW. It's mentioned in the
>User's Guide.
>
> John
>
>On Thu, Mar 23, 2006 at 10:30:53PM +0100, Jan Saam wrote:
>
>
>>Hi all,
>>
>>I've noticed this behaviour in several cases, too. It seems to be a
>>limitation of the algorithm due to Kabsch 1976. For small sets of atoms
>>(~<100) the rotation sometimes is just wrong.
>>
>>Does anybody know a better alignment algorithm?
>>
>>Also it would be interesting to know if the same problem also occurs in
>>Pymol or other viewers. If it works in Pymol it would be interesting to
>>see what they do better.
>>I stumbled over a reference in the internet where someone implemented
>>the Kabsch algorithm as a Pymol script. I wonder if that one aligns your
>>molecules correctly?
>>
>>IS THERE A PYMOL USER LISTENING?
>>Could you try it with Visvaldas example?
>>
>>Regards,
>>Jan
>>
>>PS.
>>That's the original paper:
>>*[Kabsch, 1976]* Kabsch, W. (1976). A solution for the best rotation to
>>relate two sets of vectors. /Acta. Crystal/, 32A:922-923.
>>
>>Did anybody read the following paper? I don't have access to it, but it
>>might contain some information about our problem:
>>*[Kabsch, 1978]* Kabsch, W. (1978). A discussion of the solution for the
>>best rotation to related two sets of vectors. /Acta. Crystal/, 34A:827-828.
>>
>>
>>
>>Visvaldas Kairys wrote:
>>
>>
>>
>>>Hi all,
>>>
>>>I am trying to use VMD to compute RMSD between selected residues, and for
>>>practical purposes I superimpose only the atoms under interest.
>>>Below is an example of molecules (in pdb format) which cause problem.
>>>
>>>molecule 1:
>>>
>>>CRYST1 1.000 1.000 1.000 90.00 90.00 90.00 P 1 1
>>>ATOM 1 CA GLU A 1 -29.843 154.588 35.594 0.00 0.00
>>>ATOM 2 CA GLU A 2 -27.292 160.642 36.936 0.00 0.00
>>>ATOM 3 CA GLU A 3 -32.648 156.854 50.794 0.00 0.00
>>>END
>>>
>>>molecule 2:
>>>
>>>CRYST1 1.000 1.000 1.000 90.00 90.00 90.00 P 1 1
>>>ATOM 1 CA GLU A 1 44.084 6.298 12.425 0.00 0.00
>>>ATOM 2 CA GLU A 2 39.046 2.420 14.373 0.00 0.00
>>>ATOM 3 CA GLU A 3 46.929 -0.875 26.932 0.00 0.00
>>>END
>>>
>>>If I open RMSD calculator extension, unselect backbone button
>>>and type "all" for atom selection, then click "Align",
>>>the molecules do not become aligned (it's pretty obvious), and "RMSD"
>>>is quite high. If I keep clicking "Align" and subsequent "RMSD",
>>>the position of atoms and RMSD keep changing.
>>>
>>>If I simply duplicate each atom in both of molecules above,
>>>yielding 6 atoms in each, the alignment and RMSD calculation proceeds
>>>correctly.
>>>
>>>I wonder if there is some formatting problem in my molecules or it is a
>>>bug.
>>>
>>>I had similar errors with two molecules consisting of 2 atoms. I didn't
>>>look
>>>carefully at other sizes of molecules, but big molecules seem to be OK.
>>>
>>>I am running VMD 1.8.4a22 on AMD64 Linux machine.
>>>
>>>Your help will be appreciated!
>>>
>>>Sincerely,
>>>
>>>Visvaldas Kairys
>>>University of Madeira
>>>Portugal
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>--
>>---------------------------
>>Jan Saam
>>Institute of Biochemistry
>>Charite Berlin
>>Monbijoustr. 2
>>10117 Berlin
>>Germany
>>
>>+49 30 450-528-446
>>saam_at_charite.de
>>
>>
>>
>
>
> 

-- 
---------------------------
Jan Saam
Institute of Biochemistry
Charite Berlin
Monbijoustr. 2
10117 Berlin
Germany
+49 30 450-528-446
saam_at_charite.de