Hello Peter,
at 3A resolution occupancy refinement is supposed to fail, I'd say.
The large B-values can make up for reduced occupancy (as you already
pointed out to) but also large flexibility.
You might try ten refinement runs in parallel with fixed occupancies
varying from 10% to 100% and check whether you can detect signs of a local
minimum at a certain occupancy - e.g. real space CC, best fit of B-values,
etc. Yet much more you can probably not do.
If you cannot come to a satisfactory solution you may place a model ligand
and describe the situation in the PDB header. That's not ideal but
probably one of the best ways to go in my opinion.
Tim
--
Tim Gruene
Institut fuer anorganische Chemie
Tammannstr. 4
D-37077 Goettingen
GPG Key ID = A46BEE1A
On Thu, 2 Apr 2009, peter hudson wrote:
Hello all
I have refined and built model a dataset of 3.0A resolution dataset. This
model is assocaied with a ligand. After final refinement and model
building i found a big blob of Fo-Fc density of around 4sigma level at
the N-terminal of the fianl model. My protein doesnot carry any tag at
the N-terminal. But, the template i have used for the molecular
replacement carrying a tag at the N-termianl, which can occupy only 1/4th
of this positive density after superpostion. My crystallisation
conditions component cannot occupy such higher level of positive density.
Since my protein binds to a ligand, i looked carefully to the positive
density and it seems very similar to the ligand density but its obscure.
Refinement after fitting the ligand leads to very high B-factor of the
ligan, which vary for various atoms from 50-90 and simultaneously
positive density goes off. I also tried to change the occupancy of the
ligand but as i reduce the occupancy the B factor comes down at the
noraml expected average B factor value but again Fo-Fc density appears in
the map over ligand. If i leave to refine the occupancy to the programme
automatically, this lead to the zero occupancy of the few atoms in the
ligand and avearge B factor remains normal. suggestions would be
appreciated.
Thanks in advance
Peter
