Hi Lijun,
once you solved and refined the structure, and you used for this the
anisotropy truncated data, please PDB deposit both: manipulated data
(the one you used for refinement the model), and the original dataset.
This is very important. One day the software will be able to handle
highly anisotropic datasets correctly, so you would not need to apply
some suspicious manipulations to it anymore.
Pavel.
On 7/21/10 4:18 PM, Lijun Liu wrote:
Ed,
Thanks for the comment! Anisotropic correction hits the problem, and
makes good
improvement!
You suggestion about one strong reflection around 2.1 Å is very
interesting.
However, one single reflection's (along c*) ripples passes along c and
give sin/cos
pattern along c, which means along c everywhere should give such a
residual map
pattern throughout the unit cell, which is not observed. As I
mentioned, only several
small regions showed such residual maxima (please do not argue with me
I also had
lost 1 or 2 big low ordered reflections, :). Also, this reflection
needs to be very strong and
should not be overload-ignored during processing. Anyway, reflections
(0 0 39-41) are
pretty weak or normal. Cheers, the great FT!!!
Lijun
On Jul 21, 2010, at 2:28 PM, Edward A. Berry wrote:
Another possible cause of this "layering" might be a single strong
"rogue" reflection at about 2.1A along the C* axis. Since hex ice
has a lattice spacing around here, a single crystal of ice (a snowflake
captured from the LN2 or growing on the crystal in the cold stream)
could result in such spots. If one of them was close enough to the
predicted spot of your crystal, it could get integrated and show up
in your data list.
Not an ice ring, you understand, but an ice diffraction spot
appearing to be a protein diffraction spot. I think this is
what Phoebe was referring to.
If the anisotropy correction doesn't give a good map, try making
a map with resolution limited to 2.3 A. If that looks ok, extend
the resolution in small steps through the 2.1A range. If you layers
suddenly returns with one of these steps, you could eliminate a thin
shell, or narrow it down to the particular spot and eliminate that.
-ed-
Lijun Liu wrote:
No. No ice ring! Sorry I meant the strongest amplitudes of (Fo-Fc) at
the 2.1 A around region in the reciprocal space, not the strongest
densities on real space images. As Pavel suggested, this is a good case
of anisotropicity. In my original email, I used "polarized" pattern to
address this---I used the wrong nomenclature to address the same thing,
I think. Lijun
Could any of those strong spots at 2.1 be ice?
---- Original message ----
Date: Tue, 20 Jul 2010 16:13:31 -0700
From: Lijun Liu <lijun....@ucsf.edu <mailto:lijun....@ucsf.edu>>
Subject: Re: [ccp4bb] Residual densities!
To: CCP4BB@JISCMAIL.AC.UK <mailto:CCP4BB@JISCMAIL.AC.UK>
Pavel,
Thanks for the quick response. I just listed out
the Fo-Fc column of the reflection files, and it
looks the strongest densities are mostly along c*
and many located in the 2.1A region. I think you
may be right.
Lijun
On Jul 20, 2010, at 4:05 PM, Pavel Afonine wrote:
Hi,
It looks like the data is highly anisotropic -
I've seen it many times. You can either apply
anisotropy correction
(http://www.doe-mbi.ucla.edu/~sawaya/anisoscale/
<http://www.doe-mbi.ucla.edu/%7Esawaya/anisoscale/>)
or use PHENIX tools to compute the maps - anyone
of both should help. The major difference between
the two is that PHENIX will not modify your
original data in any way. Let me if using PHENIX
map calculation tools does not help.
Cheers,
Pavel.
On 7/20/10 3:19 PM, Lijun Liu wrote:
Hi all,
I solved a structure and the refinement is close
to complete. However, some residual density
puzzles me.
The SG is P212121, and the c is 84.8 Å. Along
c, in several small but not all regions, there
are some layered, alternately appeared positive
and negative residual density pieces (c is the
norm of these plane-like pieces; 3-sigma contour
for pos and neg difference densities). The
distance between the neighboring positive pieces
is measured to be ~2.1 Å, corresponding to
~1/40 *c. In one of such region, there is an
indole ring of Trp that is approximately
parallel to the plane-like residual density
pieces (attached figure). In two other regions,
there are Leu or else but not ring structure.
The data is complete to 1.9 Å. Reflection
check did not show any special imcompleteness
(ice ring loss, wedge loss, other resolution bin
loss, etc). It looks like to me that there is a
systematic loss of information which made it is
specially sensitive to ~2.1 Å reflections
(especially those along c*). It does not look
like to me to be radiation damage. Crystal was
needle and showed somehow polarized (but never
serious) diffraction pattern along c*.
Attached is a piece of map with the
above-mentioned Trp included. I would like to
know if some one also met something like this
before. Thanks for any comments.
Lijun Liu
Cardiovascular Research Institute
University of California, San Francisco
1700 4th Street, Box 2532
San Francisco, CA 94158
Phone: (415)514-2836
Lijun Liu
Cardiovascular Research Institute
University of California, San Francisco
1700 4th Street, Box 2532
San Francisco, CA 94158
Phone: (415)514-2836
Lijun Liu
Cardiovascular Research Institute
University of California, San Francisco
1700 4th Street, Box 2532
San Francisco, CA 94158
Phone: (415)514-2836
Lijun Liu
Cardiovascular Research Institute
University of California, San Francisco
1700 4th Street, Box 2532
San Francisco, CA 94158
Phone: (415)514-2836
