From: Jacob Keller
Date: 1 September 2011 21:03
Dear Crystallographers,
I recently have been working with a 2.5 Ang SeMet peak wavelength
dataset which contains 2 cys's and also a couple of bona fide Cl ions
(reasonable b-factor/site is semi-buried/water does not work). In the
FFT anomalous difference map using PhiC from the refined model and
Dano, I can see the MSE's at ~10 sigma, but no Cl ions, even though Cl
should have f" = ~0.3 versus Se's f" = ~4, and no S's in the cys,
despite f" = 0.23e. There is really no anomalous peak at all--is it
just the smallness of the signal, or are the Se's somehow "swamping
out" the other signal? Perhaps the phases are tainted by the presence
of semet in the model?
Looking for suggestions,
Jacob Keller
*******************************************
Jacob Pearson Keller
Northwestern University
Medical Scientist Training Program
*******************************************
----------
From: Bosch, Juergen
Where in refinement of your model are you ?
At an early stage I wouldn't be surprised to only see SeMets but once you've refined your structure and go back to calculate an anomalous map with the improved phases you might double your signal for SeMet and start seeing sulfurs.
An alternative explanation, you've blasted your crystals at the synchrotron and the remaining anomalous signal is too weak to show the sulfurs.
Just two thoughts,
Jürgen
......................
Jürgen Bosch
Johns Hopkins University
Bloomberg School of Public Health
Department of Biochemistry & Molecular Biology
Johns Hopkins Malaria Research Institute
615 North Wolfe Street, W8708
Baltimore, MD 21205
----------
From: jens Preben Morth
Hi Jacob
I agree with Juergen, and just add that your Cys and Cl might not be fully occupied.
cheers
Preben
J. Preben Morth, Ph.D
Group Leader
Membrane Transport Group
Nordic EMBL Partnership
Centre for Molecular Medicine Norway (NCMM)
University of Oslo
P.O.Box 1137 Blindern
0318 Oslo, Norway
http://www.jpmorth.dk
----------
From: Jacob Keller
Update:
I tried more anomalous maps, this time with the originally-deposited
data at 1.8 Ang (mine were similar, substrate-soaked crystals) and
phases from the refined model, and the Se sites are now ~40-50 sigma,
and there is still totally nothing at the Cl and S sites, even though
in 2Fo-Fc the Cl is ~9 sigma, and the S is 8 sigma (the Se is ~15
sigma). If it has reasonably-high electron density, shouldn't it have
at least some anomalous scattering? I am wondering whether somehow the
model phases are biasing the map, but I can't really imagine how that
would be...
JPK
----------
From: Pete Meyer
I'd be surprised if the model phases were introducing bias into an anomalous difference map (they might be adding noise, but that's another story). I've also never seen feedback in model-phased anomalous difference maps (experimental-phased anomalous difference maps can have pretty bad feedback).
Have you tried cutting the known Se sites out of your map? Or phaser sas residual maps?
Pete
----------
From: Randy Read
Dear Jacob,
The signal for weak anomalous sites can be stronger in Phaser SAD LLG maps than in model-phased anomalous difference Fouriers, especially if the substructure already contains the stronger sites, so that you're just looking for what is still left to be explained in the SAD data. You could try running Phaser in MR-SAD mode, giving the current protein model as a partial structure, providing the substructure of the Se sites, and looking for purely-imaginary scatterers ("LLGCOMPLETE SCATTERER AX" in a script, or check the box labelled "Complete with purely anomalous scatterer" in the ccp4i GUI). In this situation you don't want to look for S or Cl atoms as such, because the real part of their scattering is already accounted for well in the protein model.
Best wishes,
Randy Read
Randy J. Read
Department of Haematology, University of Cambridge
Cambridge Institute for Medical Research
Wellcome Trust/MRC Building
Hills Road
Cambridge CB2 0XY, U.K.
----------
From: James Holton
It is quite possible that the S- and Cl- signal is being lost under that from the Se sites. Could be:
1) simply noise that would swamp the S-SAD signal anyway
2) you just aren't contouring your map low enough ("sigma" is not on an absolute scale)
3) trigonometry. Remember, with SAD you are not looking at the heavy atom contribution (FH) directly, you are looking at the projection of FH that is orthogonal to FP (the protein contribution). Hence the weaker atom positions have less and less influence on DANO as the Se signal becomes stronger.
One way to test the last hypothesis is to calculate anomalous differences from your model and see what that anomalous-difference map looks like. You can get DANOcalc for a given PDB file and wavelength using the CCP4 Suite and my script:
http://bl831.als.lbl.gov/~jamesh/mlfsom/ano_sfall.com
or you can do it with phenix.fmodel after you look up the f', f" for all your elements at the wavelength of interest.
It is interesting to see what happens if you set f" = 100 electrons or more. If f" is too big, then it swamps FP, and you can no longer get phases by SAD.
-James Holton
MAD Scientist
----------
From: Gerard Bricogne
Dear Jacob,
... or you could use the "anomalous residual map" in SHARP, the first
program to offer this kind of calculation 15 years ago or so: see
La Fortelle, E. de & Bricogne, G. (1997). Methods Enzymol. 276, 472–494.
and/or the SHARP manual at
http://www.globalphasing.com/sharp/manual/chapter5.html#RESIDMAPS
We have routinely observed since the inception of these maps that their
signal-to-noise ratio is significantly greater than that of the ordinary
anomalous difference Fourier maps. The same holds for the isomorphous or
dispersive residual maps vs. ordinary difference Fourier maps.
If you do not find the expected signal, it could just be that your data
collection protocol gave rise to too much differential radiation damage on
average between the measurement of Bijvoet pairs. This would tend to happen
especially if your crystal has low symmetry.
Good luck!
Gerard.
--
===============================================================
* *
* Gerard Bricogne
* *
* Global Phasing Ltd. *
* Sheraton House, Castle Park
* Cambridge CB3 0AX, UK
* *
===============================================================
----------
From: Eleanor Dodson
To repeat James's points
1) DANO (or Diso) maps are are equivalent to calculating a projection - even for perfect data the term is F" sin(PhiP-PhiA) so
it wont be as clear as a LLG map.. from either SHARP or Phaser .
2) the SIGMA level will reflect the Se signal strength
I usually calculate a DANO map and do a peak search to check whether there is a weak signal. (GUI FFT anomalous - peak search - then read those "coordinates" into coot to compare to the existing model. The weak scatterers are usually there but often way down the list..
3) Of course if the anom diff is a bit damaged by radiation or something else your signal may be buried by the noise.
Eleanor
from) so the Se sites. Could be:
No comments:
Post a Comment