Hi all -
On 15 Jul 2009, at 7:03, Lijun Liu wrote:
Hello dear Wei,
1) Your dataset has a high overall Rmerge. The outmost shell (70%)
is very high, which suggests a need to shrink resolution. What
about I/s(I), redundancy and completeness? Also, how many
reflections (percentage) have been subjected to rejection? Too many
rejections may mean a SG error.
70% Rmerge alone in the outer shell, can indeed be reasonable for such
high symmetry, depending on e.g. redundancy and what Lijun also
mentions. I would be more interested in the low resolution Rmerge to
be able to have an opinion about the dataset ...
2) Software analyses including Phenix.xtriage give good suggestions,
but they should be used reasonably. P6n22 could be twinned from P6n
or P3n12 or P3n21 or even P3n, when close to be perfectly twinned.
Correct, but phenix.xtriage as well as Pointless, will still pick up
possible twining even if you have merged data to P6n. Still, its best
to integrate the data to P1, which is always perfectly legitimate, and
then run one of these programs to get the likelihood of the correct
space group, and consider twining.
3) Please try a little more analyses of the cell shape and A.U.
shape (very long in your case), with Matthews' coefficient analyses,
you may get more hints about packing and SG.
Indeed, for example keep in mind that for 2.9 A resolution from a good
beamline, you might have a solvent content of 70%, quite easily.
4) Try to find one molecule first in such a big A.U. [Automatic
scripts have a higher chance to fail than a smart manual scrip (very
personal)].
Here I do disagree, since I found that automated scripts are
explicitly more useful in the case of multiple copies, and smarter
than (my) old manual scripts. The reason is that they are likely to
keep a lot of initial solutions for the first molecule, and then look
for the second based on these many possibilities for the first, and so
on, which can be very powerful.
5) Use a lower resolution ~4-5 Å data to do a better self rotation
function. Why do not paste a .ps plot for the SR peaks.
Using low resolution for SF is a good idea, 5-6 is also nice. But
please do not send us any attachments at the BB ;-)
A.
Good luck.
Lijun
On Jul 14, 2009, at 4:23 AM, Wei Zhang wrote:
Dears,
I am doing molecular replacement of a protein complex with a P622
data set with large cell parameters (a=b=135, c=480). The data set
seems well. R merge is 0.17 for all and 0.70 for the last shell of
2.9 angstrom. I am not sure it is a complex in the crystal. Phenix
analysis reveal there is no twin. The proposed protein complex is
about 70 kDa with a larger subunit of 50 kDa and a small subunit of
20 kDa. The matthews analysis indicates that there might be 3
complexes in the ASU. The structure of the 50 kDa subunit is known
while the 20 kDa one is unknown. But molecular replacement failed
with either Phaser or Molrep.
Self-rotation with CNS reported the result as below:
! index, psi, phi, kappa, RF-function ( 0.25)
1 0.000 0.000 180.000 29.7217
Self-ratation with Molrep reported the result as below:
Number of RF peaks : 30
theta phi chi alpha beta gamma
Rf Rf/sigma
Sol_RF 1 0.00 0.00 0.00 0.00 0.00 0.00
0.3444E+05 17.13
Sol_RF 2 90.00 -80.07 179.99 0.00 180.00 -19.86
5061. 2.52
Sol_RF 3 90.00 -65.53 179.99 0.00 180.00 -48.94
4890. 2.43
Sol_RF 4 90.00 -76.12 179.99 0.00 180.00 -27.76
4722. 2.35
Sol_RF 5 11.42 90.00 61.00 30.00 11.53 30.00
2438. 1.21
Sol_RF 6 164.38 60.00 179.99 60.00 -31.24 120.00
1850. 0.92
Sol_RF 7 85.13 -141.54 179.97 38.32 -170.26 141.40
1805. 0.90
Sol_RF 8 90.00 -60.00 90.00 30.00 -90.00 -30.00
1764. 0.88
Sol_RF 9 84.07 -144.87 179.96 34.98 -168.15 144.73
1743. 0.87
Sol_RF 10 72.25 -60.00 89.30 46.76 -84.03 -13.24
1665. 0.83
Sol_RF 11 138.24 -130.77 180.00 49.23 83.52 130.77
1608. 0.80
Sol_RF 12 170.36 30.00 179.99 30.00 -19.28 150.00
1590. 0.79
Sol_RF 13 82.04 60.00 179.99 60.00 -164.07 120.00
1571. 0.78
Sol_RF 14 141.01 30.00 179.99 30.00 -77.99 150.00
1554. 0.77
Sol_RF 15 123.63 -155.38 179.98 24.62 112.74 155.38
1517. 0.75
Sol_RF 16 148.80 30.00 179.99 30.00 -62.39 150.00
1450. 0.72
Sol_RF 17 142.61 30.00 180.00 30.00 -74.78 150.00
1439. 0.72
Sol_RF 18 72.25 -150.00 179.99 30.00 -144.50 150.00
1422. 0.71
Sol_RF 19 47.42 -120.00 179.98 60.00 -94.83 120.00
1417. 0.71
Sol_RF 20 153.12 30.00 179.98 30.00 -53.75 150.00
1292. 0.64
Sol_RF 21 155.99 30.00 179.98 30.00 -48.01 150.00
1274. 0.63
Sol_RF 22 138.05 30.00 179.99 30.00 -83.91 150.00
1258. 0.63
Sol_RF 23 54.12 120.00 89.15 60.00 69.32 0.00
1254. 0.62
Sol_RF 24 129.04 -136.48 179.99 43.52 101.93 136.48
1193. 0.59
Sol_RF 25 138.65 -164.29 179.99 15.71 82.69 164.29
1185. 0.59
Sol_RF 26 84.16 51.37 179.98 51.37 -168.32 128.63
1176. 0.59
Sol_RF 27 90.00 150.00 59.82 60.00 59.82 -60.00
1162. 0.58
Sol_RF 28 43.66 -133.74 179.98 46.26 -87.31 133.74
1161. 0.58
Sol_RF 29 127.85 -169.33 179.99 10.67 104.29 169.33
1153. 0.57
Sol_RF 30 124.15 30.00 179.99 30.00 -111.71 150.00
1147. 0.57
My question is:
1. For a 70 kDa protein compelx, is it common to have such a large
cell with a dimention as long as 480 angstrom?
2. Is it possible that the longest dimention of cell is doubled? If
it is, how to divide it?
3. How to interpret the self-rotation results. The results from CNS
and Molrep differs so much.
4. Any other suggestions on the molecular replacement are
appraciated.
Thanks.
Wei Zhang
PKU
Lijun Liu, PhD
Institute of Molecular Biology
Department of Physics
University of Oregon
Eugene, OR 97403
541-346-5176
http://www.uoregon.edu/~liulj/
