RMS fit summary

From: samantha hughes <s.j.hughes_at_ic.ac.uk>
Date: Wed 27 Oct 1999 16:51:02 +0100

Hi

Thanks to all who responded to my query regarding RMS analysis of
two structures with different numbers of atoms (ie a native and
mutant protein), especially to Bill Ross. I will summarise here:

It is not possible to do a conventional RMS comparison between
structures with differing numbers of atoms, as the RMS analysis
is on an atom-by-atom basis ie.
 atom #1, structure 1 is compared to atom #1 ,
structure 2
      #2, structure 1 is compared to atom #2 ,
structure 2
      . .
      . .
      . .
 atom #200, structure 1 is compared to atom #200,
structure 2

Thus, if you make a change in the structure 2 that adds or
removes an atom compared to structure 1 , your RMS values will be
garbage from that point on. You will end up for example
comparing a CA to a N instead of another CA. This holds true
even if you "only" select backbone atoms for the fit, or any
other group for that matter.

To make comparisons between structures with slightly different
atom numbers, you need to change the crd/rst/pdb file of one to
match the other and use the same prmtop for both in CARNAL. This
can be done by removing atoms with the PTRAJ facility in AMBER. I
don't know of any way to add atoms other than manually. The atom
sequence will need to be renumbered from the point where you add
the atoms.

Thanks again for your help.

Samantha
Received on Wed Oct 27 1999 - 08:51:02 PDT
Custom Search