Hello,
What about the enantiomorphic space groups (the 4(3) screw axes instead
of the 4(1) screw axes) ? These cannot be distinguised on the basis of
structure factor amplitudes (unless you have an anomalous scatterer) nor
on the basis of the specific extinctions (both screw axes extinguish the
same reflections).
However molecular replacement distinguishes between the two. I did not
see any mention of these enantiomorphic space groups in your post hence
my asking... Phaser allows to test everything. The CCP4 program
Pointless is also good at letting you know what is the most likely space
group.
Hence you may have a combination of wrong space group and twinning. I
can't really say with the information provided. I have had one case
(also molecular replacement) where I had to test every single space
group (a hexagonal lattice) in order to hit the correct one (that was
before the days of Phaser).
Also, with molecular replacement the first thing I do is to compare the
initial R-f / R-free to the R-factors that are expected for a random
distribution of atoms in the asymmetric unit (from memory, ca. 0.6).
Once you have refined, the numbers go down so you can't really say
anything any more.
Furthermore, sometimes it is necessary to carry out the molecular
replacement searches with a smaller set of atoms (for example when you
are expecting to find a tetramer you carry out the searches with a
dimer, or with a monomer - and sometimes you have the surprise to
discover that the solvent content of your crystal is quite high and that
the tetramer is formed by 2 dimers that are related by a
crystallographic 2-fold axis - or the arrangement of molecules in your
multimer is different from that in the search model, to sum it up: in
crystallography, everything goes!).
HTH,
Fred.
On 01/12/12 01:59, ruisher hu wrote:
Hi, Dear CCP4 group,
I recently collect one dataset and indexed as P4 space group. When I
try to do MR with a tetramer as input, I found the solution file
suggested P41.
SOLU SET RFZ=5.2 TFZ=10.3 PAK=0 LLG=239 LLG=366 TFZ==20.9
SOLU SPAC P 41
SOLU 6DIM ENSE ensemble1 EULER 50.265 0.217 219.800 FRAC 0.50029
0.49916 -0.00408 BFAC -0.04000
SOLU ENSE ensemble1 VRMS 0.639
SOLU SET RFZ=5.2 TFZ=10.0 PAK=0 LLG=239 TFZ==11.2 LLG=365
SOLU SPAC P 41
SOLU 6DIM ENSE ensemble1 EULER 128.607 179.813 38.677 FRAC 0.49991
0.49905 -0.00208 BFAC -0.06264
SOLU ENSE ensemble1 VRMS 0.642
However when I did refinement in phenix, I have some trouble getting
the R/Rfree down. It complains about some twinning or maybe higher
symmetry like P422. When I apply twin law, h,-k,-l, I'm able to refine
to 0.34/0.37 but still, hard to continue the refinement. The strategy
I used for refinement is xyz coordinates, real-space, individual
B-factors, occupancies, with NCS restraints and twin law.
So I tried to rescaled into P41212 space group, and run MR again, with
two chains as an input and found one solution
SOLU SET RFZ=4.5 TFZ=9.5 PAK=0 LLG=115 TFZ==10.0 LLG=118 TFZ==10.4
SOLU SPAC P 41 21 2
SOLU 6DIM ENSE ET EULER 234.6 0.7 305.0 <tel:234.6%200.7%20305.0>
FRAC 1.00 0.49 -0.17 BFAC -0.14
Ensemble ET RMS variance(s): 0.96
However, when i tried to refine, the Rfree is high as 0.51 at the
first round. Does this mean this is not the right solution or maybe
some problems with the space group?Any suggestions for next step?
Thanks very much.
--
Fred. Vellieux (B.Sc., Ph.D., hdr)
IBS / ELMA
41 rue Jules Horowitz
F-38027 Grenoble Cedex 01
Tel: +33 438789605
Fax: +33 438785494
