Re: [ccp4bb]: refinement with higher resolution data

[Igor Mochalkin <igor@msi.com>]

***  For details on how to be removed from this list visit the  ***
***    CCP4 home page http://www.dl.ac.uk/CCP/CCP4/main.html    ***

Hi Guan,

>dear all:
>This is a refinement problem with higher resolutiuon data.
>Once the 2.4 A structure  was solved, and now I have collected 1.4 A data for
>the same crystal form. The cell constants differ very little. The sequence is
>not known, but some homologous sequences exist. Two molecules in the A.U., but
>I have not found the NCS aixs.

Since you have solved the isomorphous structure at 2.4 A using NCS, you 
know the position of your dimer. I would suggest that you use one of your 
two molecules (the most complete one) and refine it independently as a 
rigid body for two molecules of this dimer using 1.4 A at low resolution 
(15 - 2.8) A or so. Then from these solutions you can construct your new 
dimer and get a new NCS matrix (get_ncs_matrices.inp).

>I use the 2.4A structure (water deleted, B factor reset 15.00) as initial 
>model
>to do MR with the new 1.4 A data. The data is very good (Rmerge is 4.0% 
>for the
>whole data set and 20% for last shell. I am using CNS to refine it.(just 
>did rigid.inp
>and anneal.inp, and the R and Rf were only lowered a little, R~35%, Rf~36%).
Just for curiosity, why did you set Bs to 15 A2? Is this is the default 
value in generate.inp of CNS?

>The question is:
>1. Most residues can be found in the map, only one loop including 9 a.a. 
>in both
>molecules (each 64 a.a.) have very poor density in the sa-omit.map.
>Someone suggests that I delete these residues from the PDB file, and add 
>waters
>in the these places in the map first, then maybe I can find some good 
>density there.
>Is this method ok? or other choice?
>How should I do in this case?
I would suggest to use density modification, plus since you also have 2 
molecule/ASU do averaging, and do not place water molecules in these places.

>2. At this high resolution (I directly use the 1.4 A data to refine), how 
>about
>the NCS restraints? Since I have not found the NCS axis for the 2 
>molecules, can
>I use NCS restrints? (esp. at this resolution)
I would suggest to start the refinement using NCS at 2.8 or 2.5  A. and 
refine it down to 2.1 A.  From this point, you might think about dropping 
NCS and refine your molecules individually.

Regards,
Igor



>3. in CNS inputs files, there are choices on B factors and bulk-slovent. 
>When shall
>I use the "anisotropic" for B factors and "true" for bulk_solvent? In the 
>example
>files of CNS tutorial, I found them were choosed in the early stages.
>4. can anyone give some advices on my refinement work? or what strategy 
>should I
>use when refine a low resolution structure with higher resolution data?
>
>Thanks in advance.
>guan

==========================================
Igor Mochalkin, PhD.		 igor@msi.com
Product Specialist, X-Ray	Tel:  (858) 799-5224
Molecular Simulations Inc.	Fax: (858) 458-0136
9685 Scranton Road		 URL: http://www.msi.com
San Diego, CA 92121-3752
+++++++++++++++++++++++++++++++++++++++++++