[phenixbb] R-free rules

Nathaniel Echols NEchols at lbl.gov
Fri Nov 13 13:20:46 PST 2009

On Nov 13, 2009, at 1:08 PM, Ralf W. Grosse-Kunstleve wrote:
>> [ phenix.refine v1.5 rel. tag 2 intel linux 2.6 ]
>> if i provide the test set .mtz (flag = "1") in refinement  
>> separately from
>> the data .mtz, it looks like reflection_file_converter.py does not  
>> flag
>> the work set as "0", if the work set lacks such a flag -- is that  
>> right?
> I think that's true.
> Nat wrote a more comprehensive tool for generating R-free flags, which
> are included in recent nightly builds. I'm pretty sure Nat generates
> the flags for all possible reflections.
>> i.e. when providing the test set, it looks like all observed data  
>> must be
>> in the test set, however the work set has flag "0".
> Hm... not sure I understand, but try Nat's new tools first. It is
> easiest through the GUI, but there's also a command-line version,
> phenix.reflection_file_editor.

I'm not sure I understand either - does this mean that the MTZ file  
with the test set only has Miller indices (i.e. h,k,l) for the  
reflections actually in the test set, and the work set is simply  
missing?  If so, I'm almost certain that my program isn't going to do  
any better, since it assumes that the R-free flags are comprehensive,  
but if you send me (not the list!) the reflection files I can probably  
fix them quickly.

Nathaniel Echols
Lawrence Berkeley Lab
NEchols at lbl.gov

More information about the phenixbb mailing list