Hi,

The second Patterson peak is twice the first (considering lattice translations, where 1 is equivalent to 0 modulo 1), and then if you triple the first vector you'll get minus the first vector (again considering lattice translations, i.e. 3/4 is equal to 1 - 1/4 which is equivalent to 
-1/4), which is equivalent by symmetry to the first vector so wouldn't appear in the peak list.  So the Patterson indicates 4 copies separated by 0, 1, 2 and 3 times the top Patterson vector, in approximately the same orientation.

We've haven't fully dealt with the complications of multiple tNCS-related copies in Phaser yet, but for this type of case there is a reasonable treatment.  You should add two commands to the Phaser job:

TNCS NMOL 4
TNCS PATT PERCENT 80

The first says that the Patterson translation is repeated 4 times, and the second will cause the second Patterson peak to be ignored.

I'd suggest repeating the Phaser run with these commands and making sure that you end up with the same solution as you got when the tNCS was ignored.  When tNCS is ignored, it's possible to end up with a solution that is only partially correct, which would be one explanation for having some molecules that look better in density than others.

Best wishes,

Randy Read

On 31 Jul 2012, at 13:11, Qixu Cai wrote:

> It's a P212121 dataset. I have used phaser to find four solution in ASU.
> 
> This is the phaser log file:
> ------------------------------------------------------------
> PEUDO-TRANSLATIONAL NCS VECTOR
> --------------------------------------------------------------
> 
>   Space Group :       P 21 21 21
>   Patterson Symmetry: P m m m
>   Resolution of All Data (Number):        2.45  49.00 (22968)
>   Resolution of Patterson (Number):       5.00   9.99 (2364)
>   There were 2 non-origin distinct peaks (i.e. more than 15 angstroms from the
>   origin)
> 
> <!--SUMMARY_BEGIN-->
>   84.1% origin:   FRAC 0.500 0.000 0.250   (ORTH   32.0    0.0   32.2)
>   72.2% origin:   FRAC 0.000 0.000 0.500   (ORTH    0.0    0.0   64.4)
> <!--SUMMARY_END-->
> 
>   More than one pseudo-translational ncs vector found
>      Correction factors will not be applied
> 
> 
> PS: I have used phenix.xtrige and found the p-value of
> pseudo-translational ncs is very little, which indicates the exist of
> the pseudo-translational ncs. And no twin found in this dataset.
> 
> Now the problem is two in the four molecules of an ASU have worse
> electron density than the other two molecules. And after rigidbody and
> restraint refinement by refmac without "twin refinement", the R/Rfree
> is a little high (0.33/0.36). And if I turn on the "twin refinement"
> in refmac, the R/Rfree is 0.30/0.33.
> 
> So, my question is, there is not twin in my data but
> pseudo-translational ncs, is it suitable to use "twin refinement" in
> refmac, which has a good R/Rfree result.
> 
> Thanks a lot for your help.
> 
> Best wishes,
> 
> Qixu Cai
> 
> 
> 2012/7/31, Eleanor Dodson <[log in to unmask]>:
>> More details - what do you mean by pesudo-translational symmetry ?
>> Are there two molecules related by a translation vector? or its it
>> something more complicated?
>> Eleanor
>> 
>> On 31 July 2012 10:47, Qixu Cai <[log in to unmask]> wrote:
>> 
>>> Dear all,
>>> 
>>> Can I use the "twin refinement" to refine the pesudo-translational
>>> symmetry dataset?
>>> 
>>> Thanks a lot for your help.
>>> 
>>> Best wishes,
>>> 
>>> Qixu Cai
>>> 
>> 
> 
> 
> -- 
> Qixu Cai
> Email: [log in to unmask]
> School of Life Sciences,
> Xiamen University, Fujian, China

------
Randy J. Read
Department of Haematology, University of Cambridge
Cambridge Institute for Medical Research      Tel: + 44 1223 336500
Wellcome Trust/MRC Building                   Fax: + 44 1223 336827
Hills Road                                    E-mail: [log in to unmask]
Cambridge CB2 0XY, U.K.                       www-structmed.cimr.cam.ac.uk