View NMR str on RasMol

B Tracy Nixon (btn1@psu.edu)
Wed, 05 Apr 1995 13:44:33 -0400

I discovered that I wanted to view some NMR structures of a
protein that I am interested in learning about. I downloaded
the full file from Brookhaven to my computer, and opened it
with RasMol. I got a structure to view, ... But, I noticed
that the file had 20 models rather than one, as alternative
solutions to the distance constraints from the data set do
exist. I wondered if I could view all twenty backbones overlain
as I have seen at seminars. After a trip to the chemistry
department silicon graphics station, I was still left wondering
how to do this - the folks there were not able to help me.
Although the program, an updated version of Frodo, imported the
data, it gave out of space errors when objects were identified.

I brought the data file into an editor (note that one of the
data files was 7 MB in size, the others ~3 MB), and after reading
the PDB description details, tried some experiments. First, I
used a macro to delete all but the CA carbons from the data
file, and renumbered the consecutive models to keep an
increasing index.

This file was successfully brought into RasMol, and even though
the opening screen was unusual, displaying as backbone had the
desired effect of overlaying all the models.

Then, I discovered that I did not need to renumber the CA atoms
for the different models, and if I selected residue 13 in the
structure, I got all 20 model versions of residue 13.

Finally, I tried simply deleting the lines ENDMDL from the data
file. This time I got all of the atom coordinates imported into
RasMol, including side chains etc. My computer is a pentium
16 MB RAM, and it works okay with these data (each model is for
125 residues). Perhaps the size of the files would make working
with such data impractical on some systems.

I note on overlaying backbones from the different models of my protein that
some parts form a tight bundle of backbones, while others form a
fairyly ill defined region. The ends of the protein are also "fuzzy"
in this manner. Does this mean that there was too little information
in the data set to define the backbone precisely, and if so does this simply
reflect a problem with NMR or does this actually identify
flexible regions of the protein?

My instincts are the former, but I am wondering about the
latter.

If anyone cares to join me in viewing these files, they are of
CheY and NtrC. Be aware that I think you will have to edit
the .pdb datafiles in order to view the multiple backbones in
RasMol. I have posted the edited versions on the Valine Group
home page, and I have posted a simple basic program to perform the
needed editing on these or other NMR.pdb files. If anyone is
interested, check out the link on the Valine Group home page
(http://millreef.mcb.psu.edu/pps/valinegr).

Tracy Nixon - Valine Group