[Chimera-users] RMSD fit of an object to atom

[Eric Pettersen]

On Mar 7, 2007, at 10:45 AM, Navnit Kumar Mishra wrote:

>
> Dear folks;
>
> Is there any simple way in chimera to do rmsd fitting of the object to
> the atom. I would like to measure the rmsd of an density peak with the
> atom. Could someone give the detail, how can I do it.

Since the RMSD between a single peak and an atom is the same as the  
distance, you can use the Volume Path Tracer tool to drop a "marker"  
on the density peak and then measure the distance from the marker to  
the atom (control click the marker [selecting it] and shift-control  
double-click the atom [selecting it also and bringing up a context  
menu] and choose "Show Distance").  I don't know if once you've  
dropped the marker near the peak if there is any way to migrate it  
automatically to the absolute peak.  Perhaps Tom Goddard will be able  
to provide advice on that.

> And one another
> thing, how can I get rid of the problem with " At least four atoms
> should be selected", I am interested to do 'match' for two atoms.

There is no unique transformation between coordinate systems unless  
you specify a non-planar set of points (i.e. at least 4).  With only  
two, you are basically specifying two axes that you want to be  
collinear.  Not only are there two ways to make them collinear, but  
there are an infinite number of relative rotations possible about the  
collinear axes.

With three points, the non-uniqueness problem is a little less  
obvious.  Consider two sets of points A-B-C and D-E-F that form right  
angles.  Clearly you can superimpose A-B-C with D-E-F in a 1-to-1  
fashion.  But you can equally validly superimpose A-B-C with F-E-D  
(i.e. 180 degree flip).  So... you need 4 points.

If you don't care about the non-uniqueness, you can use the "align"  
command to place both pairs of atoms on the Z axis, with one atom of  
each pair superimposed.

--Eric

                         Eric Pettersen
                         UCSF Computer Graphics Lab
                         pett at cgl.ucsf.edu
                         http://www.cgl.ucsf.edu