[Chimera-users] independent rotation mode

Fri May 9 18:36:39 PDT 2003

Date: Fri, 9 May 2003 13:57:27 -0500 (CDT)
From: Steve Ludtke <stevel at blake.3dem.bioch.bcm.tmc.edu>
Reply-To: sludtke at bcm.tmc.edu
To: Thomas Goddard <goddard at cgl.ucsf.edu>

Hi again. I just ran across a bug I thought I'd report. The session file 
is rather large, so I won't attach it initially. Let me know if you have 
problems reproducing it.  Anyway, I have loaded several volume data sets 
which represent different versions of the same structure. Then:

1) set 'independent' rotation mode
2) set one model so it is active, but not shown (in the model panel)
3) rotate the models
4) set the model in 2 to be shown again

Even though it was active, the model in 2 was not rotated.  I have a 
complex scenario of models where I need to do this operation frequently, 
so right now all my rotations keep getting messed up unless I make sure 
all the models are active AND shown before rotating anything.
Cheers...

PS - This happens with both the current 1700 version and the previous one.
-- 
----------------------------------------------------------------------------
Steven Ludtke, PhD              |        Baylor College of Medicine
sludtke at bcm.tmc.edu             |               Co-Director
stevel at alumni.caltech.edu       | National Center For Macromolecular Imaging 
V: (713)798-6989                |    Dept of Biochemistry and Mol. Biol.
instant messenger: sludtke42    |            * Those who do ARE *
http://ncmi.bcm.tmc.edu/~stevel |           The converse also holds

Date: Fri, 9 May 2003 15:04:20 -0500 (CDT)
From: Steve Ludtke <stevel at blake.3dem.bioch.bcm.tmc.edu>
Reply-To: sludtke at bcm.tmc.edu
To: Thomas Goddard <goddard at cgl.ucsf.edu>

Not THAT big a rush. Whenever you have a chance    :^)

Anyway, I was trying to avoid having to explain the whole situation, but 
here goes. I have a PDB model, a density map based on the PDB model and 3 
different EM maps of the same molecule (each a little different). All are 
aligned and centered. What I want to do is have the EM map(s) with 
overlaid PDB structure on the left side of the screen and the x-ray map 
with overlaid structure on the right half of the screen. Then I use 
'independent mode' to rotate everything about its own center so I can see 
how the EM map and x-ray map compare. As long as everything is active and 
displayed, it works beautifully (except the PDB model center seems very 
slightly off, so you see a very small amount of the centering artifact you 
described). Anyway, on the 'EM' side, I have 3 models overlaid. Generally 
I want to turn their display on and off as I rotate each one around and 
look at them. However, even though all 3 remain active, when I turn off 
the display of a model (ONLY in independent mode), it no longer gets 
rotated properly.   Hope that helps clarify...
cheers.

-- 
----------------------------------------------------------------------------
Steven Ludtke, PhD              |        Baylor College of Medicine
sludtke at bcm.tmc.edu             |               Co-Director
stevel at alumni.caltech.edu       | National Center For Macromolecular Imaging 
V: (713)798-6989                |    Dept of Biochemistry and Mol. Biol.
instant messenger: sludtke42    |            * Those who do ARE *
http://ncmi.bcm.tmc.edu/~stevel |           The converse also holds

From: Tom Goddard

The problem is that indepedent rotation mode only rotates about the
the center of the displayed part of the model.  If nothing is displayed
there is no center to rotate about.

Date: Fri, 9 May 2003 20:02:39 -0500 (CDT)
From: Steve Ludtke <stevel at blake.3dem.bioch.bcm.tmc.edu>
Reply-To: sludtke at bcm.tmc.edu
To: Thomas Goddard <goddard at cgl.ucsf.edu>

I understand. I sort of figured it was something like this. Well, I'll 
give you my perspective as a 'user' from the EM community. As funny as it 
seems, I would actually want the opposite of the behavior you stated. That 
is, the molecules would always rotate about their 'true' center even if 
part wasn't displayed. The best solution from my perspective would either 
A) have a different 'independent' mode where things always rotated about 
their true centers or 
B) Make it so models that aren't displayed at all use their true center. 
This doesn't seem quirky to me. If the model is active, you assume some 
sort of rotation will occur. Presumably if part of the molecule is 
visible, the whole thing still gets rotated, just about a different 
center. So, it's odd for it not to get rotated at all.

>From a user perspective, a problem I keep running into is that I 
accidentally do a rotation when not all of the molecules are active. This 
produces a relative rotation that is virtually impossible to undo through 
the GUI. Translations aren't a problem. They're pretty easy to 'undo' 
almost perfectly. But rotations are really tricky. When that happens I end 
up having to revert to my last saved session. So, having the model not 
rotate at all just because it isn't visible is doubly annoying...

Anyway, just my perspective...

-- 
----------------------------------------------------------------------------
Steven Ludtke, PhD              |        Baylor College of Medicine
sludtke at bcm.tmc.edu             |               Co-Director
stevel at alumni.caltech.edu       | National Center For Macromolecular Imaging 
V: (713)798-6989                |    Dept of Biochemistry and Mol. Biol.
instant messenger: sludtke42    |            * Those who do ARE *
http://ncmi.bcm.tmc.edu/~stevel |           The converse also holds

Hi Steve,

  I agree it is odd for the undisplayed model not to be rotated at all.
It seems reasonable that when the model is undisplayed it should
use the center of the whole model if you are using independent rotation
mode.

  There is the additional problem that the centers of the models
don't coincide.  When you display an isosurface, it uses the center
of the bounding box that just contains the surface.  It does not
use the actual bounds of the whole density map.

  I see that Chimera has support for fixing the center of rotation for
each model.  This could be used to get exactly the rotation behaviour
you want.  But we need a user interface to let you say which models
you want to group together with a common center.  I could write something
like this in Python.  This would address both the undisplayed models
not rotating, and the not quite right centers of rotation.

  I realize that accidentally unaligning two models in Chimera is
a disaster -- it usually means starting over.  I think we need a rotation
and translation undo to address this.

    Tom