[phenixbb] possible Xtriage bug??

Michael Thompson miket at chem.ucla.edu
Tue Jun 25 15:37:48 PDT 2013


Hello Developers,

I would like to report a possible bug in newer versions of phenix.xtriage. I have noticed recently that I am getting some strange results for the results of the L-test when I run xtriage, as described below: 

My data show a strong non-crystallographic translation, described by the vector <uvw>=(1/3, 2/3, 1/2). This NCS translation is readily identified by xtriage. When Xtriage performs the L-test, the output states:


 L test for acentric data

 using difference vectors (dh,dk,dl) of the form:
(2hp,2kp,2lp)
  where hp, kp, and lp are random signed integers such that
  2 <= |dh| + |dk| + |dl| <= 8

  Mean |L|   :0.504  (untwinned: 0.500; perfect twin: 0.375)
  Mean  L^2  :0.340  (untwinned: 0.333; perfect twin: 0.200)

These results don't make sense, as the statistics show an "anti-twinned" effect, which should not happen with the L-test. It appears that xtriage is using reflections that do not have the same expected intensity values due to the NCS translation. 

The strange thing is that I have another xtriage logfile, which I produced in September of 2012 using the same input data, that indicates the reflections used for the L-test were chosen "using difference vectors (dh,dk,dl) of the form: (3hp,3kp,2lp)." This choice of reflections does account for proper parity, and I get results that are much more sensible (Mean |L|=0.394; Mean L^2=0.221). 

I am wondering why I am getting different results now than I did last September? Has something changed in the underlying code, or am I missing something silly? I do not have an easy way of determining which version of phenix I used when the L-test was conducted using sensible reflections (the xtriage logfile doesn't contain this info as far as I can tell), but the version that seems to choose reflections inappropriately is 1.8.2-1334.

Thanks a lot for the help, and for phenix being awesome in general!

Mike






-- 
Michael C. Thompson

Graduate Student

Biochemistry & Molecular Biology Division

Department of Chemistry & Biochemistry

University of California, Los Angeles

miket at chem.ucla.edu


More information about the phenixbb mailing list