*** For details on how to be removed from this list visit the ***
*** CCP4 home page http://www.ccp4.ac.uk ***
Andreas,
If your crystal really does include a dimer formed by crystallographic
twofold symmetry, and you search with the appropriate dimer, you should
get an enhanced signal in the rotation function compared to searching
just with the monomer, and the translation for a correct solution will
not be impaired. Packing will obviously be poor.
I suppose one thing that packing functions in automated MR programs
might do is treat cases where the distance between centres of gravity of
symmetry-related molecules = zero as a good thing if there is internal
symmetry in the search model, rather than rejecting them.
I haven't done it myself, but I assume that with phaser you should be
able to extract the monomer from your 'correct' dimer position, and fix
that, before running a translation function on the rest of your rotation
peaks. If the position really is correct, then phaser might be able to
find a solution for additional molecules
Cheers,
Charlie
[EMAIL PROTECTED] wrote:
*** For details on how to be removed from this list visit the ***
*** CCP4 home page http://www.ccp4.ac.uk ***
Dear molecular replacement cracks out there,
I'm working on a non-trivial molecular replacement problem (bad model,
poor data, many molecules in au) and am hitting the wall. This might be
because the problem cannot be solved or because I'm blind to the solution.
Run on a list of normal mode-perturbed models, Molrep didn't find obvious
solutions but suggested a dimer I should use as model instead of the
monomer. This makes good sense, as my protein is like dimeric (though I
can't see NCS in the data). What's the basis of this suggestion, Alexei?
Using the dimer didn't help Molrep find the solution, but Phaser suggested
a rotation function (Z-score = 3) whose 26 associated translation
functions have Z-scores between 7 and 9. All solutions are rejected
because of clashes (~100).
Forcing the transformations I can write out 26 potential solutions and
look at them in PyMOL. They are lined up along the a axis of the unit
cell (c222_1). The a axis is also coincident with the two-fold that
relates the two monomers of the dimer (thus the nearly complete overlap
indicated by 100 clashes). One monomer is symmetry-related to the other.
Why does Phaser put the dimer on a symmetry axis? Would this not give
wrong solutions in any case where you search with a dimer? Is it possible
(and sensible) to define the a axis in C222_1 as a dyad axis for molecular
replacement?
At first I thought my problems were due to misindexing, but running phaser
or Molrep with data reindexed in all other possible spacegroups (p1, p2,
p2_1, c2, c222, see statistics at the end of the email) didn't give
solutions (nor high Z-scores in Phaser). For example, in P1, Molrep
places 4 dimers but the contrast values decrease with every placed dimer
(from 5 to 2). Does that tell me that I'm completely off?
Next I was thinking twinning but there is no indication for that (Yeates,
sfcheck).
Currently I'm running Molrep with the dyad option and half of the
suggested dimer (i.e. the initial monomeric model). In C222 and C222_1
Molrep comes up with a solution containing four models (as expected from
Matthews coefficient) and good contrast values (11 and 120 for C222; 92
and 85 for C222_1). Looking at the solutions, I don't see dimers. The
other space groups are work in progress...
Furthermore, I plan to do the dyad search with a fixed model running
Molrep 26 times with all solutions from Phaser.
Am I chasing waterfalls, or does it sound like I'm up to something? Any
comments or suggestions are highly appreciated. And, if you got that far,
thanks for reading through all this ;-)
Andreas
And yes, I have whipped myself and a student for the last six months
trying to get better crystals, derivatives, data, etc.
Indexing in different space groups - output from XDS for signal/noise >=
0, all data:
RESOLUTION NUMBER OF REFLECTIONS COMPLETENESS R-FACTOR
LIMIT OBSERVED UNIQUE POSSIBLE OF DATA observed
P1 a total 105246 62621 108903 57.5% 8.2%
P1 b total 105224 62606 108903 57.5% 8.2%
P2 total 106172 50601 56092 90.2% 9.0%
P2_1 total 106164 50594 56081 90.2% 9.0%
C2 a total 108603 38721 55096 70.3% 10.4%
C2 b total 107130 48446 55231 87.7% 9.2%
C222_1 total 110472 27435 28860 95.1% 11.1%
C222 total 110479 27441 28871 95.0% 11.1%
R-FACTOR COMPARED I/SIGMA R-meas Rmrgd-F S_norm/
expected S_ano
P1 a 8.3% 85250 7.52 11.7% 26.4% 0.00
P1 b 8.4% 85236 7.51 11.7% 26.4% 0.00
P2 9.1% 102652 8.35 12.3% 26.6% 1.12
P2_1 9.1% 102650 8.35 12.3% 26.6% 1.12
C2 a 10.6% 102377 8.75 12.5% 23.5% 1.06
C2 b 9.3% 101481 8.34 12.2% 25.4% 1.08
C222_1 11.4% 110384 10.50 12.9% 19.9% 1.02
C222 11.4% 110386 10.50 12.9% 19.9% 1.02
.
--
Charlie Bond
Professorial Fellow
University of Western Australia
School of Biomedical, Biomolecular and Chemical Sciences
M310
35 Stirling Highway
Crawley WA 6009
Australia
[EMAIL PROTECTED]
+61 8 6488 4406
