Hi Kama,
Maybe this will gives some ideas of things to do if the problem is bad
initial MR phases.
Sascha Keller et al. Acta Cryst. (2006). D62, 1564–1570
They report averaging with solvent flattening in DM using just the data
to 5 A and slow phase extension from there to make big difference.
Good luck,
David
--
David Aragao, Ph. D.
Postdoctoral Researcher
Membrane Structural and Functional Biology group
CES department, University of Limerick
Ireland
[log in to unmask] wrote:
> Dear all,
>
> Please anyone give me any advice on the problem below.
>
> Diffraction spots not so bad, collected by HKL2000 to 2.2A. Cell parameter is
> a=99.958, b=99.958, c=197.237, Judged by clear systematic absence l=6n and no
> mm on hk0, space group P61 or P65, that expects two molecules of 411aa in AU.
> Rsym = 7.1, redundancy = ~11. Strangely, plots of reciprocal space show strong
> diffractions on h, k, 7n plane, and significantly weak ones or no diffraction
> on h, k, 7n+1 and h, k, 7n-1 plane. Of course there are some exceptions, but
> this diffraction pattern goes to high resolution with ~35 A2 of Wilson B
> factor.
>
> Cumulative intensity distribution plot of TRUNCATE shows acentric observation
> curve is much higher than theoretical, and centric observation is lower (half
> of theoretical)(fig). This phenomena is quite similar seen in this ccp4bb, for
> example, recent case in Qiang Chen wrote at April 3 2008
>
> I started CNS molecular search using a know structure with 40 % identity,
> found two solution structure which are not overlap each other. These are
> related to pseudo-two fold axis parallel to c. Both makes 61 screw axis along
> to c, forming like a double helix.
>
> Molrep found same solution and also showed pseudo-translation vectors.
> Peak 1: trans.vector /ort/ : 0.000 0.000 84.717
> trans.vector /frac/: 0.000 0.000 0.430
> <= 3/7
> Peak 2: trans.vector /ort/ : 0.000 0.000 27.783
> trans.vector /frac/: 0.000 0.000 0.141
> <= 1/7
> Peak 3: trans.vector /ort/ : 0.000 0.000 4.895
> trans.vector /frac/: 0.000 0.000 0.025
>
> I made model structures created by modeller and started refinement by CNS and
> refmac. 2fofc map clearly showed density for 70% of the model. But additional
> model building and refinement process could not drop R/Rfree (43/48) any more.
> Messy and chopped densities (>3 sig.) are still lying neighbor to the
> molecules in fofc map. Some of the density are traceable as a strand and a
> helix. But about 20% of the density are crushing to the refined models.
>
> I think that this is a bad case of pseudo-translation generated 3 vectors X 2
> molecules along c* axis, but this is not twining. Is my recognition correct?
> Please give any suggestion how to overcome this situation by handling this
> data, or controlling crystal growth.
> Any suggestions will be appreciated.
>
> Kama
>
> ------------------------------------
> Katsuhiko Kamada, Ph.D.
>
> Senior Research Scientist of
> Chromosome Dynamics Laboratory
> RIKEN Discovery Research Institute
>
> Bioscience building # N-255,
> 2-1 Hirosawa, Wako, Saitama 351-0198
> JAPAN
> E-mail: [log in to unmask]
> Phone: +81-48-467-9532
> Fax: +81-48-462-4673
> ------------------------------------
>
|