It is important that your new smaller test set include only reflections
that were in your previous test set, and not include any of the previous
working reflections.
By now you probably have several replies detailing how to achieve this in ccp4.
If you happen to be using CNS, there is an easy way to pair down the free set:
Use the make_cv.inp script. Use a cv file with the old free set as input.
{===>} reflection_infile="oldfile.cv";
The old free set will probably be labeled "test". Make a new column
to hold the 2.5% flags:
{===>} test_set="test2p5";
Select the new set from the old set:
{* Additional selection for generation of test set *}
{===>} ref_sel=( test=1 );
select 50% of the reflections:
{===>} percentage=50.0;{50% of 5% is 2.5%}
and give a new name for the new file
{===>} reflection_outfile="newfile.cv";
I think those are the salient points, but if it doesn't work
email me and I'll send a working version
HTH,
ed
(formerly [log in to unmask])
Jason C Porta wrote:
> Hi all,
>
> I am currently refining a high-resolution structure that has many
> reflections (~180,000). I would like to halve the R-free set from 5% to
> 2.5%, and am unsure how to do so. Any advice will be greatly appreciated.
>
> Jason Porta
> Graduate Student
> Dept. Biochemistry and Molecular Biology
> University of Nebraska Medical Center
> Omaha, NE 68198
|