[phenixbb] problem with a twinned data set

Pascal EGEA pascal at msg.ucsf.edu
Sat Mar 1 16:04:39 PST 2008

Hi All

This a question of  general Crystallography and use of Phenix to deal  
with twinned data.

I am having difficulties with a data set. We have data (2.6 ang  
resolution) that can be scaled and reduced in P222 with a Rmerge of  
7%, the systematic absences show a screw axis ( it is P22(1)2).
However MR in Phaser failed in this orthorhombic settings when  
searching for  2 molecules of the  complex in the AU.
If we reduce and scale the data in P2(1) and look at the diffraction  
pattern in HKLVIEW there is  a /mmm symmetry. In the monoclinic P2(1)  
setting we can find 4 molecules of the complex (by MR in Phaser) and  
can refine it with Phenix  to a Rfac/Rfree of 28%/34%. 2 molecules  
have good electron density whereas the two other ones have one of  
their domain very poorly defined in density. This is looking very  
suspicious to me and I am wondering if this refined structure is  
partially wrong?

Meanwhile  I used phenix.triage on the data processed in P222 and I  
am confused with the output.

##                   Twinning Analyses                ##

Using data between 10.00 to 3.36 Angstrom.

Determining possible twin laws.

The following twin laws have been found:

| Type | Axis   | R metric (%) | delta (le Page) | delta (Lebedev) |  
Twin law |
|  PM  | 4-fold | 2.851        | 0.000           | 0.013           | - 
l,k,h   |
M:  Merohedral twin law
PM: Pseudomerohedral twin law

   0 merohedral twin operators found
   1 pseudo-merohedral twin operators found
In total,   1 twin operator were found

  Analysing possible twin law :  -l,k,h

Results of the H-test on acentric data:

  (Only 50.0% of the strongest twin pairs were used)

mean |H| : 0.368   (0.50: untwinned; 0.0: 50% twinned)
mean H^2 : 0.194   (0.33: untwinned; 0.0: 50% twinned)
Estimation of twin fraction via mean |H|: 0.132
Estimation of twin fraction via cum. dist. of H: 0.115

Britton analyses

   Extrapolation performed on  0.14 < alpha < 0.495
   Estimated twin fraction: 0.127
   Correlation: 0.9955

By comparison if I run the detect_twin routine of CNS it tells me  
that they are no merohedral twin laws for the point group 222 (using  
the statistical method of Yeates) ? This is  confusing me quite a bit.

Assuming that the twinning law suggested by Phenix is correct how  
should I proceed?
I have noticed the section concerning refinement using twinned data  
in Phenix
phenix.refine data.hkl model.pdb twin_law="-k,-h,-l"  

but It seems I  only have a MR solution in P2(1) but not in P222(1)  
so how can I refine in P222(1)

I will greatly appreciate your input, many thanks.

Pascal F. Egea, PhD
University of California San Francisco
Department of Biophysics and Biochemistry
Robert Stroud Laboratory
pascal at msg.ucsf.edu

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://phenix-online.org/pipermail/phenixbb/attachments/20080301/bc094212/attachment-0003.htm>

More information about the phenixbb mailing list