Hi Feng,
just my speculations (of course, the respective authors of the program
you use may provide you with a better answer):
In terms of refinement results, all pairs of numbers below look the same
to me (regardless the refinement program used):
0.192/0.250
0.191/0.247
0.202/0.254
0.194/0.255
that is, they are within "expected" deviations. Typically, a refinement
program does anisotropic scaling and bulk-solvent correction before
computing the R-factors. This step numerically is very sensitive to
almost everything, so it's not surprising at all that you observe these
small differences between different versions of the program.
Pavel.
On 1/4/10 9:35 AM, Feng Guo wrote:
Hi,
I have a question about the refinement by Refmac program in CCP4 package. I
tried both the old version(1.4.4) and the latest version (2.0.5) for restrained
refinement, and they gave me different result. For old version, it's
0.192/0.250--> 0.191/0.247, while for the latest version, R/Rfree=0.202/0.254-
->0.194/0.255. For both version, I use the weighting term of 0.05. I am
wondering which version will be better? or I did something wrong? Thank you.
Feng
