Arghhh...R-factor woes...CNS refinement

maneesh maneesh at drunkenbastards.com
Fri Jan 24 09:50:09 EST 2003


Hi all,
I am part of (an apparently growing) group of people who knows just
enough crystallography to be dangerous.

I have a set of structures of 4 tetrameric (33 residue/monomer) 4
helix bundles which I have collected in house data at a pretty good
resolution (~1.9 - 2.1 A).

I use molrep from ccp4 (which works really great for me BTW), and then
I like to use that solution to give to CNS for refinment.  My problem
is that after running:

rigid.inp

bgroup.inp
minimize.inp
bindividual.inp
minimize.inp
....

my R factor gets stuck at around 30%.  The R-free is generally a few
percent behind (34%), and follows closely during the minimization.  I
load up maps into X-fit and fix some gross errors in side chains, but
my factors don't improve, the electron density maps however look quite
correct.  I tried adding the solvent at this stage, but there was only
a tiny drop (~1%) in the R-factor.

I tried simulated annealing an in the first set of minimizations
(looking at the inp file, is it just adjusting B-factors at the first
stage?), the R-factor drops like a rock to where I would think it
should be (just a little over 20%); but then as it continues, by the
end the Rfree is around 40% and the R factor hovers around 30%.  Can
anyone tell me what is going on?

I obtain the data from Crystal Clear and the statistics seem fine (low
Rmerge, and I generally cut the data off at I/sig ~= 1.9-2.1).  Doing
the Laue checks confirm that we have the right space group....(we have
mroe evidence that the space group is correct from similar peptides
which made the same kinds of crystals solved with Xe diffusion)...

Can anyone give me some ideas as to what to try to get the R-factor
down, I am reasonably confident that I have good quality data...

I am sure these grips are common, but I'd really appreciate any
suggestions.





More information about the Xtal-log mailing list