Hi Faisal,
nobody can answer the question based on the information you gave, but _you_ can answer it in the following simple way:
a) preparation: from your 3.2A reflection file master.mtz, copy the reflections to 3.5A to cut35.mtz, and the reflections to 3.8A to cut38.mtz. This can be done with CCP4 cad or sftools.
b) calculation: take your 3.2A model and refine it for zero cycles against cut35.mtz , just to obtain R/Rfree for the unchanged 3.2A model against the 3.5A data. "Refining for zero cycles" might sound strange but it ensures that you always use the same program for R/Rfree calculation.
c) comparison: compare the R/Rfree values against what you got for the 3.5A model against the same data: which has the lower Rfree, and lower Rfree-Rwork gap?
d) choice: the 3.2A model is better (worse) if it gives a lower (higher) Rfree against the 3.5A data than the 3.5A model does. It should also give the lower (higher) gap.
If you find out that the 3.2A model is better, you might want to verify your finding against the 3.8A data, just to be sure.
If you find out that the 3.5A model is better, you might want to repeat the above procedure, substituting files and models accordingly.
This is called "paired refinement technique". In my experience, often the answers are quite clear. Sometimes the answers are contradictory; in those cases I often find that the model can be improved, and I later obtain consistent answers with the better model.
You should also check R/Rfree in the highest resolution bins; those numbers should not be much higher than 40% (unless your data are twinned or you have pseudo-translation; in those cases your reflection statistics are different).
HTH,
Kay
On Fri, 8 Jan 2016 19:29:54 +0530, khaja faisal tarique <[log in to unmask]> wrote:
>Hi everyone.
>
>I have a 3.2A data with a solvent content of 67%. Although the I/I SIG
>value is around 2 at this resolution the r and r free is 28 and 36. As the
>resolution is cut to 3.5A, r and r free becomes 26 and 32 and i / isig
>becomes 4. Lastly at 3.8 it becomes 24 and 28 with I / isig value around 6
>The electron densities for many side chains are missing in each data set
>may be because of high solvent content. My question is which data set is
>optimum for deposition in pdb or reporting in a paper keeping in view the r
>and rfree value. There is hardly any difference in the quality of electron
>density in each of three data sets.
>
>Regards
>
>Faisal
>
|