Re: [ccp4bb] Space group problem

[Tim Gruene]

Hello Rain Field,

I third Felix and Craig, there is nothing unusual about the statistics
table you present. The data look pretty good and I would assume
P6_something and integrate further than 3.1A.

Regards,
Tim

On 05/07/2014 08:16 PM,  wrote:
> additional info:
> If I let xds go through, it will choose P6. actually pointless suggest 
> P62/P64.
> The thing is the Rmeas and Rmerge are significantly higher for P2/P3/P6 than 
> P1, especially the highest shell.
> That indicates those higher symmetry ones are not the choice, it that right?
> (Actually, this is also the reviewer's question) 
> 
> 
> P6 log from xds:
> 
>  SUBSET OF INTENSITY DATA WITH SIGNAL/NOISE >= -3.0 AS FUNCTION OF RESOLUTION
>  RESOLUTION NUMBER OF REFLECTIONSCOMPLETENESS R-FACTOR  R-FACTOR 
> COMPARED I/SIGMA   R-meas  CC(1/2)  Anomal  SigAno   Nano
>LIMIT OBSERVED  UNIQUE  POSSIBLE OF DATA   observed  expected  
> Corr
> 
>  8.454512 216   239   90.4%   3.1%  3.3% 
> 4511   95.55  3.2%   100.0*41*   1.124 182
>  6.277208 330   330  100.0%   4.0%  4.0% 
> 7208   72.47  4.1%   100.0*22*   1.211 293
>  5.208925 401   405   99.0%   5.4%  5.4% 
> 8925   56.60  5.5%99.9*120.965 368
>  4.55   10585 470   474   99.2%   6.3%  6.5%
> 10585   48.76  6.5%   100.0* 30.841 436
>  4.09   11898 529   542   97.6%   9.1%  9.0%
> 11898   37.80  9.3%99.9*-60.809 495
>  3.75   12425 550   573   96.0%  16.9% 16.7%
> 12425   22.09 17.3%99.7* 10.855 519
>  3.48   14344 638   639   99.8%  26.3% 26.1%
> 14344   14.57 26.9%99.3*-30.783 601
>  3.26   15001 668   671   99.6%  53.6% 56.1%
> 150017.15 54.8%96.9*-20.745 633
>  3.08   14177 704   725   97.1% 114.8%124.6%
> 141652.95117.7%87.5*-30.632 637
> total   990754506  4598   98.0%   9.3%  9.5%
> 99062   30.73  9.5%   100.0* 10.8334164
> 

-- 
Dr Tim Gruene
Institut fuer anorganische Chemie
Tammannstr. 4
D-37077 Goettingen

GPG Key ID = A46BEE1A



signature.asc
Description: OpenPGP digital signature

Re: [ccp4bb] Space group problem

[Craig Bingman]

It is definitely P6 or P3 with a twinning about the 
unique axis.  The differences in merging statistics between P2 P3 and P6 are 
not very significant in my opinion.  

On May 7, 2014, at 1:16 PM,   wrote:

> additional info:
> If I let xds go through, it will choose P6. actually pointless suggest 
> P62/P64.
> The thing is the Rmeas and Rmerge are significantly higher for P2/P3/P6 than 
> P1, especially the highest shell.
> That indicates those higher symmetry ones are not the choice, it that right?
> (Actually, this is also the reviewer's question) 
> 
> 
> P6 log from xds:
> 
> SUBSET OF INTENSITY DATA WITH SIGNAL/NOISE >= -3.0 AS FUNCTION OF RESOLUTION
> RESOLUTION NUMBER OF REFLECTIONSCOMPLETENESS R-FACTOR  R-FACTOR 
> COMPARED I/SIGMA   R-meas  CC(1/2)  Anomal  SigAno   Nano
>   LIMIT OBSERVED  UNIQUE  POSSIBLE OF DATA   observed  expected   
>Corr
> 
> 8.454512 216   239   90.4%   3.1%  3.3% 
> 4511   95.55  3.2%   100.0*41*   1.124 182
> 6.277208 330   330  100.0%   4.0%  4.0% 
> 7208   72.47  4.1%   100.0*22*   1.211 293
> 5.208925 401   405   99.0%   5.4%  5.4% 
> 8925   56.60  5.5%99.9*120.965 368
> 4.55   10585 470   474   99.2%   6.3%  6.5%
> 10585   48.76  6.5%   100.0* 30.841 436
> 4.09   11898 529   542   97.6%   9.1%  9.0%
> 11898   37.80  9.3%99.9*-60.809 495
> 3.75   12425 550   573   96.0%  16.9% 16.7%
> 12425   22.09 17.3%99.7* 10.855 519
> 3.48   14344 638   639   99.8%  26.3% 26.1%
> 14344   14.57 26.9%99.3*-30.783 601
> 3.26   15001 668   671   99.6%  53.6% 56.1%
> 150017.15 54.8%96.9*-20.745 633
> 3.08   14177 704   725   97.1% 114.8%124.6%
> 141652.95117.7%87.5*-30.632 637
>total   990754506  4598   98.0%   9.3%  9.5%
> 99062   30.73  9.5%   100.0* 10.8334164

Re: [ccp4bb] Space group problem

I thought there was a theoretical Rmeas number for I/sigma=2 that is around 
50%? In my case, the Rmeas is much higher. 
Before, we reported P62(4) and got rejected. 
I'll just change a journal.  Hopefully reviewer will not be too picky about 
that. 
Thanks to everyone!

Re: [ccp4bb] Space group problem

[Felix Frolow]

Your symmetry is of family P6.
There is no dramatic difference between 7.8% and 9.3% of Rmerge
Go with pointless, wave reviewer  nonsense.
 BTW reviewer of what journal he is?

Dr Felix Frolow   
Professor of Structural Biology and Biotechnology, Department of Molecular 
Microbiology and Biotechnology
Tel Aviv University 69978, Israel

Acta Crystallographica F, co-editor

e-mail: mbfro...@post.tau.ac.il
Tel:  ++972-3640-8723
Fax: ++972-3640-9407
Cellular: 0547 459 608

On May 7, 2014, at 21:16 ,
wrote:

> additional info:
> If I let xds go through, it will choose P6. actually pointless suggest 
> P62/P64.
> The thing is the Rmeas and Rmerge are significantly higher for P2/P3/P6 than 
> P1, especially the highest shell.
> That indicates those higher symmetry ones are not the choice, it that right?
> (Actually, this is also the reviewer's question) 
> 
> 
> P6 log from xds:
> 
> SUBSET OF INTENSITY DATA WITH SIGNAL/NOISE >= -3.0 AS FUNCTION OF RESOLUTION
> RESOLUTION NUMBER OF REFLECTIONSCOMPLETENESS R-FACTOR  R-FACTOR 
> COMPARED I/SIGMA   R-meas  CC(1/2)  Anomal  SigAno   Nano
>   LIMIT OBSERVED  UNIQUE  POSSIBLE OF DATA   observed  expected   
>Corr
> 
> 8.454512 216   239   90.4%   3.1%  3.3% 
> 4511   95.55  3.2%   100.0*41*   1.124 182
> 6.277208 330   330  100.0%   4.0%  4.0% 
> 7208   72.47  4.1%   100.0*22*   1.211 293
> 5.208925 401   405   99.0%   5.4%  5.4% 
> 8925   56.60  5.5%99.9*120.965 368
> 4.55   10585 470   474   99.2%   6.3%  6.5%
> 10585   48.76  6.5%   100.0* 30.841 436
> 4.09   11898 529   542   97.6%   9.1%  9.0%
> 11898   37.80  9.3%99.9*-60.809 495
> 3.75   12425 550   573   96.0%  16.9% 16.7%
> 12425   22.09 17.3%99.7* 10.855 519
> 3.48   14344 638   639   99.8%  26.3% 26.1%
> 14344   14.57 26.9%99.3*-30.783 601
> 3.26   15001 668   671   99.6%  53.6% 56.1%
> 150017.15 54.8%96.9*-20.745 633
> 3.08   14177 704   725   97.1% 114.8%124.6%
> 141652.95117.7%87.5*-30.632 637
>total   990754506  4598   98.0%   9.3%  9.5%
> 99062   30.73  9.5%   100.0* 10.8334164

Re: [ccp4bb] Space group problem

additional info:
If I let xds go through, it will choose P6. actually pointless suggest P62/P64.
The thing is the Rmeas and Rmerge are significantly higher for P2/P3/P6 than 
P1, especially the highest shell.
That indicates those higher symmetry ones are not the choice, it that right?
(Actually, this is also the reviewer's question) 


P6 log from xds:

 SUBSET OF INTENSITY DATA WITH SIGNAL/NOISE >= -3.0 AS FUNCTION OF RESOLUTION
 RESOLUTION NUMBER OF REFLECTIONSCOMPLETENESS R-FACTOR  R-FACTOR 
COMPARED I/SIGMA   R-meas  CC(1/2)  Anomal  SigAno   Nano
   LIMIT OBSERVED  UNIQUE  POSSIBLE OF DATA   observed  expected
  Corr

 8.454512 216   239   90.4%   3.1%  3.3% 
4511   95.55  3.2%   100.0*41*   1.124 182
 6.277208 330   330  100.0%   4.0%  4.0% 
7208   72.47  4.1%   100.0*22*   1.211 293
 5.208925 401   405   99.0%   5.4%  5.4% 
8925   56.60  5.5%99.9*120.965 368
 4.55   10585 470   474   99.2%   6.3%  6.5%
10585   48.76  6.5%   100.0* 30.841 436
 4.09   11898 529   542   97.6%   9.1%  9.0%
11898   37.80  9.3%99.9*-60.809 495
 3.75   12425 550   573   96.0%  16.9% 16.7%
12425   22.09 17.3%99.7* 10.855 519
 3.48   14344 638   639   99.8%  26.3% 26.1%
14344   14.57 26.9%99.3*-30.783 601
 3.26   15001 668   671   99.6%  53.6% 56.1%
150017.15 54.8%96.9*-20.745 633
 3.08   14177 704   725   97.1% 114.8%124.6%
141652.95117.7%87.5*-30.632 637
total   990754506  4598   98.0%   9.3%  9.5%
99062   30.73  9.5%   100.0* 10.8334164

[ccp4bb] Space group problem

Hi all,
I have a 360 degree data set collect on home beam.
I used XDS to integrate the frames in P1. 
I progressively merge the data from P1 to P2 or P1 to P3 in XDS and attach the 
log below.
The cell looks like P3 and pointless suggest P6. But the Rmerge and Rmeas are 
much higher than normal at I/sigmaI=2. 
I think P1 might be the true space group. But the Rpim reported by aimless 
seems high in the high resolution shell. Why is that?
Thanks!

  LATTICE-  BRAVAIS-   QUALITY  UNIT CELL CONSTANTS (ANGSTROEM & DEGREES)
REINDEXING TRANSFORMATION
 CHARACTER  LATTICE OF FIT  a  b  c   alpha  beta gamma

 *  31aP  0.0  62.5   82.5   82.6  60.1  89.9  90.0   -1  0 
 0  0  0 -1 -1  0  0 -1  0  0
 *  44aP  0.2  62.5   82.5   82.6 119.9  90.1  90.01  0 
 0  0  0  1  1  0  0 -1  0  0
 *  41mC  0.6 143.1   82.5   62.5  90.0  90.1  90.00  1 
-1  0  0 -1 -1  0 -1  0  0  0
 *  30mC  0.7  82.5  143.1   62.5  89.9  90.0  90.00 -1 
-1  0  0  1 -1  0  1  0  0  0
 *  35mP  1.0  82.5   62.5   82.6  90.1 119.9  90.00 -1 
-1  0 -1  0  0  0  0  1  0  0
 *  40oC  1.2  82.5  143.1   62.5  89.9  90.0  90.00 -1 
-1  0  0 -1  1  0 -1  0  0  0
 *  20mC  2.6 142.9   82.6   62.5  90.0  90.0  90.10 -2 
-1  0  0  0 -1  0  1  0  0  0
 *  23oC  3.3  82.6  142.9   62.5  90.0  90.0  89.90  0 
 1  0  0 -2 -1  0  1  0  0  0
 *  25mC  3.3  82.6  142.9   62.5  90.0  90.0  89.90  0 
 1  0  0 -2 -1  0  1  0  0  0
 *  22hP  3.5  82.5   82.6   62.5  90.1  90.0 119.90  1 
 1  0  0 -1  0  0  1  0  0  0
37mC249.8 176.5   62.5   82.5  90.0 117.8  69.31 -2 
 0  0  1  0  0  0  0  1  1  0
42oI250.0  62.5   82.5  156.1  90.0 113.6  90.0   -1  0 
 0  0  0 -1 -1  0  1 -1  1  0
39mC250.6 176.4   62.5   82.6  90.0 117.9  69.21 -2 
-2  0  1  0  0  0  0  0  1  0
33mP434.8  62.5   82.5   82.6 119.9  90.1  90.01  0 
 0  0  0  1  1  0  0 -1  0  0
34mP435.2  62.5   82.6   82.5 119.9  90.0  90.1   -1  0 
 0  0  0  1  0  0  0 -1 -1  0
32oP435.4  62.5   82.5   82.6 119.9  90.1  90.01  0 
 0  0  0  1  1  0  0 -1  0  0
21tP437.6  82.5   82.6   62.5  90.1  90.0 119.90  1 
 1  0  0 -1  0  0  1  0  0  0



P1:
 SUBSET OF INTENSITY DATA WITH SIGNAL/NOISE >= -3.0 AS FUNCTION OF RESOLUTION
 RESOLUTION NUMBER OF REFLECTIONSCOMPLETENESS R-FACTOR  R-FACTOR 
COMPARED I/SIGMA   R-meas  CC(1/2)  Anomal  SigAno   Nano
   LIMIT OBSERVED  UNIQUE  POSSIBLE OF DATA   observed  expected
  Corr

 8.4545001165  1288   90.5%   2.2%  2.4% 
4498   51.19  2.6%99.9*100.8381030
 6.2771741844  1872   98.5%   3.1%  3.0% 
7173   35.92  3.6%99.9*11*   0.9061671
 5.2090092288  2324   98.5%   4.5%  4.3% 
9008   26.88  5.2%99.8* 40.8352168
 4.55   105972686  2744   97.9%   5.2%  5.2%
10595   22.65  6.0%99.7*-20.7762557
 4.09   120513051  3147   96.9%   7.5%  7.5%
12048   16.93  8.7%99.4* 20.7792915
 3.75   127403226  3342   96.5%  14.8% 14.5%
127389.37 17.1%98.2* 10.7603078
 3.48   143443631  3738   97.1%  22.2% 22.2%
143426.36 25.7%95.8* 00.7853471
 3.26   150793813  3948   96.6%  47.3% 48.5%
150762.99 54.8%83.7*-10.7143655
 3.08   140883797  4242   89.5%  99.7%105.6%
139451.30116.0%58.6*-20.6363133
total   99582   25501 26645   95.7%   7.7%  7.9%
99423   14.49  9.0%99.9* 10.765   23678

P2:
 SUBSET OF INTENSITY DATA WITH SIGNAL/NOISE >= -3.0 AS FUNCTION OF RESOLUTION
 RESOLUTION NUMBER OF REFLECTIONSCOMPLETENESS R-FACTOR  R-FACTOR 
COMPARED I/SIGMA   R-meas  CC(1/2)  Anomal  SigAno   Nano
   LIMIT OBSERVED  UNIQUE  POSSIBLE OF DATA   observed  expected
  Corr

 8.454499 638   711   89.7%   2.5%  2.7% 
4497   63.58  2.8%   100.0*16*   0.892 534
 6.277172 986   993   99.3%   3.5%  3.5% 
7172   46.06  3.7%99.9*100.948 872
 5.2089631202  1213   99.1%   4.9%  4.9% 
8963   35.19  5.3%99.9* 20.8491101
 4.55   105981413  1427   99.0%   5.8%  5.9%
10598   29.7

Re: [ccp4bb] AW: [ccp4bb] Refining Metal Ion Occupancy

[Tim Gruene]

-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1

Hi Ed,

I guess it depends on what 'successfully' means. The large correlation
implies quite an error margin, and I find an occupancy range of +/-
0.2 - 0.3 quite a lot, although this impression may vary depending on
what you are discussing w.r.t. the refined occupancy.

Best,
Tim

On 05/07/2014 06:01 PM, Edward A. Berry wrote:
> On 05/07/2014 10:52 AM, Tim Gruene wrote:
>> At 2.5A resolution (the resolution this thread is about)
> 
> But maps ae not made "at 2.5 A" but from say 30A to 2.5 A. In
> principle (i.e. if the 0,0,0 reflection were used), the effect of
> diminishing the amplitude of high resolution terms is not to 
> decrease the total electron density but to smooth the features: If
> the electron density profile going through the atom were a 
> rectangular box, the sharp corners would be rounded resulting in 
> less density inside the box and more outside, with the integrated 
> electron density the same. Which is exactly what you would expect 
> for an atom with increased rms displacement.
> 
> Now when the 0,0,0 reflection is absent so that the map is
> "floating" with average value zero, and also the object is smaller
> than the bragg spacing of the reflection, the 2.5 A reflection may
> contribute to total electron density - I'm not sure.
> 
> Still the main effect of increasing the B factor should be to 
> spread out the density, while decreasing occupancy reduces the
> total electron density without affecting the shape. I can readily
> imagine that refinement programs can successfully deconvolute the
> two. 90% correlation may be manageable.
> 
> eab
> 
> 
> On 05/07/2014 10:52 AM, Tim Gruene wrote: Dear Bernhard,
> 
> I just happen to collect the correlation between ADP and occupancy
> for a publication I am involved in.
> 
> At 1.5A (!) resolution, the correlation for a single ion between
> both figures is greater than 90% - there is certainly not a clear 
> difference between these factors.
> 
> One of the reasons might actually be visualised from the URL you 
> posted: At 2.5A resolution (the resolution this thread is about)
> the number of electrons for Zn with B=30 drops from 30 to 25, which
> is not so great a difference, at 1.5A it drops to about 20, which
> is still not so great a difference, i.e. the B-factor weight is not
> too far off from being constant at 'protein' resolution ranges.
> 
> Best, Tim
> 
> On 05/07/2014 02:58 PM, Bernhard Rupp wrote:
> the negative difference density surrounding your metal ion
> shows that the lower occupancy could not be fudged by a
> higher B-factor
 
 Because there is a clear difference between high B-factor and
 low occupancy: High B factor attenuates high resolution
 scattering most, while lower occupancy just evenly scales the
 scattering curve down. Ergo, the FT - the Electron density -
 also looks different, with a low occupancy causing a WIDER
 scattering curve than a comparable high B-factor, thus
 transforming into a NARROWER peak compared to high B-factor.
 
 So, you could adjust (within physically meaningful limits) B
 and n to 'reshape' the electron density. If you have a
 negative difference density 'ring', your 'observed' density
 there is less than the model density, and by reducing n you
 could reduce the wings of the model electron density peak,
 thus achieving a better match.
 
 There is also the possibility that you have - perhaps in
 addition - some truncation ripples, which are most prominent
 around heavy atoms.
 
 Figures 9-6 and 9-5 BMC. This app allows to generate the
 different scattering curve shapes, and a similar app lets you
 FT it. 
 http://www.ruppweb.org/new_comp/scattering_factors.htm
 
 
 Best, BR
 
 -Original Message- From: CCP4 bulletin board 
 [mailto:CCP4BB@JISCMAIL.AC.UK] On Behalf Of 
 herman.schreu...@sanofi.com Sent: Mittwoch, 7. Mai 2014 14:25
 To: CCP4BB@JISCMAIL.AC.UK Subject: [ccp4bb] AW: [ccp4bb]
 Refining Metal Ion Occupancy
 
 Dear Chris,
 
 In my experience, modern refinement program manage quite well
 to deconvolute occupancy and B-factor. In your case the
 negative difference density surrounding your metal ion shows
 that the lower occupancy could not be fudged by a higher
 B-factor. I would just refine occupancy and B-factor at the
 same time and let the refinement program do the
 deconvolution. If your density maps would still indicate
 problems, you always can try to manually deconvolute.
 
 By the way, your formulation >>> negative density> sounds like some cheap trick, when in fact
 you try to get a model that more accurately reflects your
 observed diffraction pattern.
 
 Best, Herman
 
 
 -Urspr�ngliche Nachricht- Von: CCP4 bulletin board 
 [mailto:CCP4BB@JISCMAIL.AC.UK] Im Auftrag von Chris Fage
>>

Re: [ccp4bb] AW: [ccp4bb] Refining Metal Ion Occupancy

[Edward A. Berry]

On 05/07/2014 10:52 AM, Tim Gruene wrote:

 At 2.5A resolution (the resolution this thread is about)


But maps ae not made "at 2.5 A" but from say 30A to 2.5 A.
In principle (i.e. if the 0,0,0 reflection were used), the effect
of diminishing the amplitude of high resolution terms is not to
decrease the total electron density but to smooth the features:
If the electron density profile going through the atom were a
rectangular box, the sharp corners would be rounded resulting in
less density inside the box and more outside, with the integrated
electron density the same. Which is exactly what you would expect
for an atom with increased rms displacement.

Now when the 0,0,0 reflection is absent so that the map is "floating"
with average value zero, and also the object is smaller than the
bragg spacing of the reflection, the 2.5 A reflection may contribute to
total electron density - I'm not sure.

Still the main effect of increasing the B factor should be to
spread out the density, while decreasing occupancy reduces
the total electron density without affecting the shape.
I can readily imagine that refinement programs can successfully
deconvolute the two. 90% correlation may be manageable.

eab


On 05/07/2014 10:52 AM, Tim Gruene wrote:

-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1

Dear Bernhard,

I just happen to collect the correlation between ADP and occupancy for
a publication I am involved in.

At 1.5A (!) resolution, the correlation for a single ion between both
figures is greater than 90% - there is certainly not a clear
difference between these factors.

One of the reasons might actually be visualised from the URL you
posted: At 2.5A resolution (the resolution this thread is about) the
number of electrons for Zn with B=30 drops from 30 to 25, which is not
so great a difference, at 1.5A it drops to about 20, which is still
not so great a difference, i.e. the B-factor weight is not too far off
from being constant at 'protein' resolution ranges.

Best,
Tim

On 05/07/2014 02:58 PM, Bernhard Rupp wrote:

the negative difference density surrounding your metal ion shows
that the lower occupancy could not be fudged by a higher
B-factor


Because there is a clear difference between high B-factor and low
occupancy: High B factor attenuates high resolution scattering
most, while lower occupancy just evenly scales the scattering curve
down. Ergo, the FT - the Electron density - also looks different,
with a low occupancy causing a WIDER scattering curve than a
comparable high B-factor, thus transforming into a NARROWER peak
compared to high B-factor.

So, you could adjust (within physically meaningful limits) B and n
to 'reshape' the electron density. If you have a negative
difference density 'ring', your 'observed' density there is less
than the model density, and by reducing n you could reduce the
wings of the model electron density peak, thus achieving a better
match.

There is also the possibility that you have - perhaps in addition -
some truncation ripples, which are most prominent around heavy
atoms.

Figures 9-6 and 9-5 BMC. This app allows to generate the different
scattering curve shapes, and a similar app lets you FT it.
http://www.ruppweb.org/new_comp/scattering_factors.htm


Best, BR

-Original Message- From: CCP4 bulletin board
[mailto:CCP4BB@JISCMAIL.AC.UK] On Behalf Of
herman.schreu...@sanofi.com Sent: Mittwoch, 7. Mai 2014 14:25 To:
CCP4BB@JISCMAIL.AC.UK Subject: [ccp4bb] AW: [ccp4bb] Refining Metal
Ion Occupancy

Dear Chris,

In my experience, modern refinement program manage quite well to
deconvolute occupancy and B-factor. In your case the negative
difference density surrounding your metal ion shows that the lower
occupancy could not be fudged by a higher B-factor. I would just
refine occupancy and B-factor at the same time and let the
refinement program do the deconvolution. If your density maps would
still indicate problems, you always can try to manually
deconvolute.

By the way, your formulation  sounds like some cheap trick, when in fact you try to get
a model that more accurately reflects your observed diffraction
pattern.

Best, Herman


-Urspr�ngliche Nachricht- Von: CCP4 bulletin board
[mailto:CCP4BB@JISCMAIL.AC.UK] Im Auftrag von Chris Fage Gesendet:
Dienstag, 6. Mai 2014 19:03 An: CCP4BB@JISCMAIL.AC.UK Betreff:
[ccp4bb] Refining Metal Ion Occupancy

Hi Everyone,

In my 2.5-angstrom structure, there is negative Fo-Fc density
surrounding a metal ion after refining in Phenix. From anomalous
diffraction I am certain of the metal's identity and position in
each monomer. Also, the ion is appropriately coordinated by nearby
side chains. Should I be refining the occupancy of the ion in
attempt to "flatten" the negative density? I am considering soaking
the metal ion into crystals or cocrystallizing and collecting
additional datasets.

Thanks for your help!

Regards, Chris



- --
- --
Dr Tim Gruene
Institut fuer anorganische Chemie
Tammannstr. 4
D-37077 Goettingen

GPG Key

Re: [ccp4bb] AW: [ccp4bb] AW: [ccp4bb] Refining Metal Ion Occupancy

[Tim Gruene]

-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1

Dear Herman,

you provided a nice pictorial explanation of the term 'correlation'.
I'll leave it to you to run the experiment - I spent a couple of weeks
on this project because the separation is quite tricky.

Apologies for being sloppy, I referred to the unit of the y-axis on
Bernhards plot, not the physical reduction of electrons.

Regards.
To,

On 05/07/2014 05:20 PM, herman.schreu...@sanofi.com wrote:
> Dear Tim,
> 
> As I see it, the issue is not how good the overall correlation is,
> but whether the differences in distribution of the electrons
> (narrow vs. broader) will show up in difference maps, which often
> show up at the edge of the ion. To make such calculations, I would
> calculate a difference map between say a model with a Zn with full
> occupancy and high B-factor vs. a model with a Zn with say 80%
> occupancy and a matching lower B-factor.
> 
> Only if you do not see any significant features is such a map, I
> will be convinced there is not a great difference in changing
> occupancy or B-factor. Besides, it is new to me that increasing the
> B-factor would reduce the number of electrons. I always assumed
> they only would get distributed differently.
> 
> Best, Herman
> 
> 
> 
> -Ursprüngliche Nachricht- Von: CCP4 bulletin board
> [mailto:CCP4BB@JISCMAIL.AC.UK] Im Auftrag von Tim Gruene Gesendet:
> Mittwoch, 7. Mai 2014 16:52 An: CCP4BB@JISCMAIL.AC.UK Betreff: Re:
> [ccp4bb] AW: [ccp4bb] Refining Metal Ion Occupancy
> 
> Dear Bernhard,
> 
> I just happen to collect the correlation between ADP and occupancy
> for a publication I am involved in.
> 
> At 1.5A (!) resolution, the correlation for a single ion between
> both figures is greater than 90% - there is certainly not a clear
> difference between these factors.
> 
> One of the reasons might actually be visualised from the URL you 
> posted: At 2.5A resolution (the resolution this thread is about)
> the number of electrons for Zn with B=30 drops from 30 to 25, which
> is not so great a difference, at 1.5A it drops to about 20, which
> is still not so great a difference, i.e. the B-factor weight is not
> too far off from being constant at 'protein' resolution ranges.
> 
> Best, Tim
> 
> On 05/07/2014 02:58 PM, Bernhard Rupp wrote:
>>> the negative difference density surrounding your metal ion
>>> shows that the lower occupancy could not be fudged by a higher
>>> B-factor
> 
>> Because there is a clear difference between high B-factor and
>> low occupancy: High B factor attenuates high resolution
>> scattering most, while lower occupancy just evenly scales the
>> scattering curve down. Ergo, the FT - the Electron density - also
>> looks different, with a low occupancy causing a WIDER scattering
>> curve than a comparable high B-factor, thus transforming into a
>> NARROWER peak compared to high B-factor.
> 
>> So, you could adjust (within physically meaningful limits) B and
>> n to 'reshape' the electron density. If you have a negative
>> difference density 'ring', your 'observed' density there is less
>> than the model density, and by reducing n you could reduce the
>> wings of the model electron density peak, thus achieving a better
>> match.
> 
>> There is also the possibility that you have - perhaps in addition
>> - some truncation ripples, which are most prominent around heavy
>> atoms.
> 
>> Figures 9-6 and 9-5 BMC. This app allows to generate the
>> different scattering curve shapes, and a similar app lets you FT
>> it. http://www.ruppweb.org/new_comp/scattering_factors.htm
> 
> 
>> Best, BR
> 
>> -Original Message- From: CCP4 bulletin board 
>> [mailto:CCP4BB@JISCMAIL.AC.UK] On Behalf Of 
>> herman.schreu...@sanofi.com Sent: Mittwoch, 7. Mai 2014 14:25
>> To: CCP4BB@JISCMAIL.AC.UK Subject: [ccp4bb] AW: [ccp4bb] Refining
>> Metal Ion Occupancy
> 
>> Dear Chris,
> 
>> In my experience, modern refinement program manage quite well to
>>  deconvolute occupancy and B-factor. In your case the negative 
>> difference density surrounding your metal ion shows that the
>> lower occupancy could not be fudged by a higher B-factor. I would
>> just refine occupancy and B-factor at the same time and let the
>> refinement program do the deconvolution. If your density maps
>> would still indicate problems, you always can try to manually
>> deconvolute.
> 
>> By the way, your formulation > density> sounds like some cheap trick, when in fact you try to
>> get a model that more accurately reflects your observed
>> diffraction pattern.
> 
>> Best, Herman
> 
> 
>> -Urspr ngliche Nachricht- Von: CCP4 bulletin board 
>> [mailto:CCP4BB@JISCMAIL.AC.UK] Im Auftrag von Chris Fage
>> Gesendet: Dienstag, 6. Mai 2014 19:03 An: CCP4BB@JISCMAIL.AC.UK
>> Betreff: [ccp4bb] Refining Metal Ion Occupancy
> 
>> Hi Everyone,
> 
>> In my 2.5-angstrom structure, there is negative Fo-Fc density 
>> surrounding a metal ion after refining in Phenix. From anomalous
>>  diffraction I am certain of the metal'

Re: [ccp4bb] Increase in Rfee as the number of cycles increase, using Refmac

[Ian Tickle]

Hi Amit

Most likely nothing is going wrong.  Your final Rfree looks quite
respectable for 2.8 Ang. data.  It's meaningless to compare the Rfree at
the beginning of a refinement with that at the end.  Rfree is only
meaningful if the refinement has converged.  So it is valid to compare the
Rfree at the end of one refinement with the Rfree at the end of another
refinement (e.g. if you changed the restraints or the weights), to decide
which starting conditions are optimal.

HTH

-- Ian


On 7 May 2014 16:11, Amit Kumar  wrote:

>
> Hello,
>
> I have a 2.8 angstrom structure of a protein ligand complex. The R/R free
> before adding the ligand was 0.21614/0.27760 and after adding the ligand,
> the R/Rfree is 0.2030/0.2691.
> .
> However while refining the structure by Refmac 5, I see a decrease in R
> factor but an increase in R free as number of cycles increase. What could
> be going wrong here?
>
> Below is the tail of the log.
>
> Ncyc Rfact Rfree FOM -LL -LLfree rmsBOND zBOND rmsANGL zANGL rmsCHIRAL $$
> $$
> 0 0.2218 0.2606 0.797 32405. 3572.4 0.0103 0.517 1.503 0.675 0.074
> 1 0.2106 0.2599 0.799 32189. 3565.2 0.0108 0.536 1.504 0.668 0.075
> 2 0.2062 0.2600 0.801 32122. 3563.5 0.0105 0.525 1.519 0.673 0.071
> 3 0.2044 0.2607 0.801 32083. 3562.7 0.0104 0.519 1.528 0.676 0.070
> 4 0.2030 0.2615 0.802 32057. 3562.9 0.0104 0.517 1.530 0.677 0.070
> 5 0.2025 0.2631 0.801 32048. 3563.9 0.0104 0.516 1.528 0.677 0.071
> 6 0.2023 0.2647 0.800 32055. 3566.1 0.0103 0.514 1.525 0.676 0.071
> 7 0.2025 0.2668 0.799 32067. 3569.0 0.0103 0.513 1.523 0.676 0.071
> 8 0.2026 0.2676 0.798 32077. 3571.1 0.0102 0.509 1.521 0.674 0.070
> 9 0.2027 0.2685 0.798 32094. 3573.8 0.0102 0.507 1.520 0.674 0.069
> 10 0.2030 0.2691 0.797 32109. 3575.6 0.0101 0.505 1.519 0.672 0.069
> $$
> $TEXT:Result: $$ Final results $$
>  InitialFinal
>   R factor 0.2218 0.2030
> R free *0.2606* *0.2691*
> Rms BondLength 0.0103 0.0101
> Rms BondAngle 1.5033 1.5193
> Rms ChirVolume 0.0735 0.0689
>
> Best,
> Amit
>

[ccp4bb] AW: [ccp4bb] AW: [ccp4bb] Refining Metal Ion Occupancy

[Herman . Schreuder]

Dear Tim,

As I see it, the issue is not how good the overall correlation is, but whether 
the differences in distribution of the electrons (narrow vs. broader) will show 
up in difference maps, which often show up at the edge of the ion. To make such 
calculations, I would calculate a difference map between say a model with a Zn 
with full occupancy and high B-factor vs. a model with a Zn with say 80% 
occupancy and a matching lower B-factor. 

Only if you do not see any significant features is such a map, I will be 
convinced there is not a great difference in changing occupancy or B-factor. 
Besides, it is new to me that increasing the B-factor would reduce the number 
of electrons. I always assumed they only would get distributed differently.

Best,
Herman



-Ursprüngliche Nachricht-
Von: CCP4 bulletin board [mailto:CCP4BB@JISCMAIL.AC.UK] Im Auftrag von Tim 
Gruene
Gesendet: Mittwoch, 7. Mai 2014 16:52
An: CCP4BB@JISCMAIL.AC.UK
Betreff: Re: [ccp4bb] AW: [ccp4bb] Refining Metal Ion Occupancy

-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1

Dear Bernhard,

I just happen to collect the correlation between ADP and occupancy for a 
publication I am involved in.

At 1.5A (!) resolution, the correlation for a single ion between both figures 
is greater than 90% - there is certainly not a clear difference between these 
factors.

One of the reasons might actually be visualised from the URL you
posted: At 2.5A resolution (the resolution this thread is about) the number of 
electrons for Zn with B=30 drops from 30 to 25, which is not so great a 
difference, at 1.5A it drops to about 20, which is still not so great a 
difference, i.e. the B-factor weight is not too far off from being constant at 
'protein' resolution ranges.

Best,
Tim

On 05/07/2014 02:58 PM, Bernhard Rupp wrote:
>> the negative difference density surrounding your metal ion shows that 
>> the lower occupancy could not be fudged by a higher B-factor
> 
> Because there is a clear difference between high B-factor and low
> occupancy: High B factor attenuates high resolution scattering most, 
> while lower occupancy just evenly scales the scattering curve down. 
> Ergo, the FT - the Electron density - also looks different, with a low 
> occupancy causing a WIDER scattering curve than a comparable high 
> B-factor, thus transforming into a NARROWER peak compared to high 
> B-factor.
> 
> So, you could adjust (within physically meaningful limits) B and n to 
> 'reshape' the electron density. If you have a negative difference 
> density 'ring', your 'observed' density there is less than the model 
> density, and by reducing n you could reduce the wings of the model 
> electron density peak, thus achieving a better match.
> 
> There is also the possibility that you have - perhaps in addition - 
> some truncation ripples, which are most prominent around heavy atoms.
> 
> Figures 9-6 and 9-5 BMC. This app allows to generate the different 
> scattering curve shapes, and a similar app lets you FT it.
> http://www.ruppweb.org/new_comp/scattering_factors.htm
> 
> 
> Best, BR
> 
> -Original Message- From: CCP4 bulletin board 
> [mailto:CCP4BB@JISCMAIL.AC.UK] On Behalf Of 
> herman.schreu...@sanofi.com Sent: Mittwoch, 7. Mai 2014 14:25 To:
> CCP4BB@JISCMAIL.AC.UK Subject: [ccp4bb] AW: [ccp4bb] Refining Metal 
> Ion Occupancy
> 
> Dear Chris,
> 
> In my experience, modern refinement program manage quite well to 
> deconvolute occupancy and B-factor. In your case the negative 
> difference density surrounding your metal ion shows that the lower 
> occupancy could not be fudged by a higher B-factor. I would just 
> refine occupancy and B-factor at the same time and let the refinement 
> program do the deconvolution. If your density maps would still 
> indicate problems, you always can try to manually deconvolute.
> 
> By the way, your formulation  density> sounds like some cheap trick, when in fact you try to get
> a model that more accurately reflects your observed diffraction 
> pattern.
> 
> Best, Herman
> 
> 
> -Urspr ngliche Nachricht- Von: CCP4 bulletin board 
> [mailto:CCP4BB@JISCMAIL.AC.UK] Im Auftrag von Chris Fage Gesendet:
> Dienstag, 6. Mai 2014 19:03 An: CCP4BB@JISCMAIL.AC.UK Betreff:
> [ccp4bb] Refining Metal Ion Occupancy
> 
> Hi Everyone,
> 
> In my 2.5-angstrom structure, there is negative Fo-Fc density 
> surrounding a metal ion after refining in Phenix. From anomalous 
> diffraction I am certain of the metal's identity and position in each 
> monomer. Also, the ion is appropriately coordinated by nearby side 
> chains. Should I be refining the occupancy of the ion in attempt to 
> "flatten" the negative density? I am considering soaking the metal ion 
> into crystals or cocrystallizing and collecting additional datasets.
> 
> Thanks for your help!
> 
> Regards, Chris
> 

- --
- --
Dr Tim Gruene
Institut fuer anorganische Chemie
Tammannstr. 4
D-37077 Goettingen

GPG Key ID = A46BEE1A

-BEGIN PGP SIGNATURE-
Version:

[ccp4bb] Increase in Rfee as the number of cycles increase, using Refmac

[Amit Kumar]

Hello,

I have a 2.8 angstrom structure of a protein ligand complex. The R/R free
before adding the ligand was 0.21614/0.27760 and after adding the ligand,
the R/Rfree is 0.2030/0.2691.
.
However while refining the structure by Refmac 5, I see a decrease in R
factor but an increase in R free as number of cycles increase. What could
be going wrong here?

Below is the tail of the log.

Ncyc Rfact Rfree FOM -LL -LLfree rmsBOND zBOND rmsANGL zANGL rmsCHIRAL $$
$$
0 0.2218 0.2606 0.797 32405. 3572.4 0.0103 0.517 1.503 0.675 0.074
1 0.2106 0.2599 0.799 32189. 3565.2 0.0108 0.536 1.504 0.668 0.075
2 0.2062 0.2600 0.801 32122. 3563.5 0.0105 0.525 1.519 0.673 0.071
3 0.2044 0.2607 0.801 32083. 3562.7 0.0104 0.519 1.528 0.676 0.070
4 0.2030 0.2615 0.802 32057. 3562.9 0.0104 0.517 1.530 0.677 0.070
5 0.2025 0.2631 0.801 32048. 3563.9 0.0104 0.516 1.528 0.677 0.071
6 0.2023 0.2647 0.800 32055. 3566.1 0.0103 0.514 1.525 0.676 0.071
7 0.2025 0.2668 0.799 32067. 3569.0 0.0103 0.513 1.523 0.676 0.071
8 0.2026 0.2676 0.798 32077. 3571.1 0.0102 0.509 1.521 0.674 0.070
9 0.2027 0.2685 0.798 32094. 3573.8 0.0102 0.507 1.520 0.674 0.069
10 0.2030 0.2691 0.797 32109. 3575.6 0.0101 0.505 1.519 0.672 0.069
$$
$TEXT:Result: $$ Final results $$
 InitialFinal
  R factor 0.2218 0.2030
R free *0.2606* *0.2691*
Rms BondLength 0.0103 0.0101
Rms BondAngle 1.5033 1.5193
Rms ChirVolume 0.0735 0.0689

Best,
Amit

[ccp4bb] LAST DAY TO REGISTER: SBGrid/NE-CAT Computing School

[Michelle Ottaviano]

We're in the final stages of planning for our June Computing School
and *registration
will close on Friday. *Join us in June for our data processing workshop
followed by Phenix and Schrodinger/BioLuminate workshops.


 *Register Here
:
Deadline Friday, May 9th*


*Quo Vadis Structural Biology 2014*
SBGrid/NE-CAT Computing School
June 5-7, 2014
Harvard Medical School -- Boston, MA


*Data Processing in Crystallography - June 5-6:*
Speakers: Kay Diederichs, Phil Evans, Zbyszek Otwinowski, Jim Pflugrath,
Dominika Borek, Piotr Sliz, Matt Benning, , Frank Murphy, David Neau,
K. Rajashankar, Nick Sauter, Jon Schuermann, Clemens Schulze-Briese

*Advances in Structure Determination with Phenix - June 7:*
Paul Adams & Nathaniel Echols, Lawrence Berkeley National Laboratory
Tom Terwilliger, Los Alamos National Laboratory


*Improved Structural Modeling Using BioLuminate - June 7: *
Melissa Landon, Ph.D., Schrodinger, Inc.

Complete meeting details can be found here:
https://sbgrid.org/news/sbgridne-cat-computing-school

Re: [ccp4bb] AW: [ccp4bb] Refining Metal Ion Occupancy

[Tim Gruene]

-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1

P.S.: For clarity, I meant to say 'calculate' (with full-matrix
refinement) instead of 'collect'.

Tim

On 05/07/2014 04:52 PM, Tim Gruene wrote:
> Dear Bernhard,
> 
> I just happen to collect the correlation between ADP and occupancy
> for a publication I am involved in.
> 
> At 1.5A (!) resolution, the correlation for a single ion between
> both figures is greater than 90% - there is certainly not a clear 
> difference between these factors.
> 
> One of the reasons might actually be visualised from the URL you 
> posted: At 2.5A resolution (the resolution this thread is about)
> the number of electrons for Zn with B=30 drops from 30 to 25, which
> is not so great a difference, at 1.5A it drops to about 20, which
> is still not so great a difference, i.e. the B-factor weight is not
> too far off from being constant at 'protein' resolution ranges.
> 
> Best, Tim
> 
> On 05/07/2014 02:58 PM, Bernhard Rupp wrote:
>>> the negative difference density surrounding your metal ion
>>> shows that the lower occupancy could not be fudged by a higher 
>>> B-factor
> 
>> Because there is a clear difference between high B-factor and
>> low occupancy: High B factor attenuates high resolution
>> scattering most, while lower occupancy just evenly scales the
>> scattering curve down. Ergo, the FT - the Electron density - also
>> looks different, with a low occupancy causing a WIDER scattering
>> curve than a comparable high B-factor, thus transforming into a
>> NARROWER peak compared to high B-factor.
> 
>> So, you could adjust (within physically meaningful limits) B and
>> n to 'reshape' the electron density. If you have a negative 
>> difference density 'ring', your 'observed' density there is less 
>> than the model density, and by reducing n you could reduce the 
>> wings of the model electron density peak, thus achieving a
>> better match.
> 
>> There is also the possibility that you have - perhaps in addition
>> - some truncation ripples, which are most prominent around heavy 
>> atoms.
> 
>> Figures 9-6 and 9-5 BMC. This app allows to generate the
>> different scattering curve shapes, and a similar app lets you FT
>> it. http://www.ruppweb.org/new_comp/scattering_factors.htm
> 
> 
>> Best, BR
> 
>> -Original Message- From: CCP4 bulletin board 
>> [mailto:CCP4BB@JISCMAIL.AC.UK] On Behalf Of 
>> herman.schreu...@sanofi.com Sent: Mittwoch, 7. Mai 2014 14:25
>> To: CCP4BB@JISCMAIL.AC.UK Subject: [ccp4bb] AW: [ccp4bb] Refining
>> Metal Ion Occupancy
> 
>> Dear Chris,
> 
>> In my experience, modern refinement program manage quite well to 
>> deconvolute occupancy and B-factor. In your case the negative 
>> difference density surrounding your metal ion shows that the
>> lower occupancy could not be fudged by a higher B-factor. I would
>> just refine occupancy and B-factor at the same time and let the 
>> refinement program do the deconvolution. If your density maps
>> would still indicate problems, you always can try to manually 
>> deconvolute.
> 
>> By the way, your formulation > density> sounds like some cheap trick, when in fact you try to
>> get a model that more accurately reflects your observed
>> diffraction pattern.
> 
>> Best, Herman
> 
> 
>> -Urspr�ngliche Nachricht- Von: CCP4 bulletin board 
>> [mailto:CCP4BB@JISCMAIL.AC.UK] Im Auftrag von Chris Fage
>> Gesendet: Dienstag, 6. Mai 2014 19:03 An: CCP4BB@JISCMAIL.AC.UK
>> Betreff: [ccp4bb] Refining Metal Ion Occupancy
> 
>> Hi Everyone,
> 
>> In my 2.5-angstrom structure, there is negative Fo-Fc density 
>> surrounding a metal ion after refining in Phenix. From anomalous 
>> diffraction I am certain of the metal's identity and position in 
>> each monomer. Also, the ion is appropriately coordinated by
>> nearby side chains. Should I be refining the occupancy of the ion
>> in attempt to "flatten" the negative density? I am considering
>> soaking the metal ion into crystals or cocrystallizing and
>> collecting additional datasets.
> 
>> Thanks for your help!
> 
>> Regards, Chris
> 
> 
> 

- -- 
- --
Dr Tim Gruene
Institut fuer anorganische Chemie
Tammannstr. 4
D-37077 Goettingen

GPG Key ID = A46BEE1A

-BEGIN PGP SIGNATURE-
Version: GnuPG v1.4.12 (GNU/Linux)
Comment: Using GnuPG with Icedove - http://www.enigmail.net/

iD8DBQFTakpIUxlJ7aRr7hoRAqZ7AKC0QbxK9jzI9wdrBgiCe9wp+fbCIwCfbQ7O
sRTdb+QcNOUvVYc4ivHbmZ8=
=kKhC
-END PGP SIGNATURE-

Re: [ccp4bb] AW: [ccp4bb] Refining Metal Ion Occupancy

[Tim Gruene]

-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1

Dear Bernhard,

I just happen to collect the correlation between ADP and occupancy for
a publication I am involved in.

At 1.5A (!) resolution, the correlation for a single ion between both
figures is greater than 90% - there is certainly not a clear
difference between these factors.

One of the reasons might actually be visualised from the URL you
posted: At 2.5A resolution (the resolution this thread is about) the
number of electrons for Zn with B=30 drops from 30 to 25, which is not
so great a difference, at 1.5A it drops to about 20, which is still
not so great a difference, i.e. the B-factor weight is not too far off
from being constant at 'protein' resolution ranges.

Best,
Tim

On 05/07/2014 02:58 PM, Bernhard Rupp wrote:
>> the negative difference density surrounding your metal ion shows
>> that the lower occupancy could not be fudged by a higher
>> B-factor
> 
> Because there is a clear difference between high B-factor and low
> occupancy: High B factor attenuates high resolution scattering
> most, while lower occupancy just evenly scales the scattering curve
> down. Ergo, the FT - the Electron density - also looks different,
> with a low occupancy causing a WIDER scattering curve than a
> comparable high B-factor, thus transforming into a NARROWER peak
> compared to high B-factor.
> 
> So, you could adjust (within physically meaningful limits) B and n
> to 'reshape' the electron density. If you have a negative
> difference density 'ring', your 'observed' density there is less
> than the model density, and by reducing n you could reduce the
> wings of the model electron density peak, thus achieving a better
> match.
> 
> There is also the possibility that you have - perhaps in addition -
> some truncation ripples, which are most prominent around heavy
> atoms.
> 
> Figures 9-6 and 9-5 BMC. This app allows to generate the different
> scattering curve shapes, and a similar app lets you FT it. 
> http://www.ruppweb.org/new_comp/scattering_factors.htm
> 
> 
> Best, BR
> 
> -Original Message- From: CCP4 bulletin board
> [mailto:CCP4BB@JISCMAIL.AC.UK] On Behalf Of
> herman.schreu...@sanofi.com Sent: Mittwoch, 7. Mai 2014 14:25 To:
> CCP4BB@JISCMAIL.AC.UK Subject: [ccp4bb] AW: [ccp4bb] Refining Metal
> Ion Occupancy
> 
> Dear Chris,
> 
> In my experience, modern refinement program manage quite well to
> deconvolute occupancy and B-factor. In your case the negative
> difference density surrounding your metal ion shows that the lower
> occupancy could not be fudged by a higher B-factor. I would just
> refine occupancy and B-factor at the same time and let the
> refinement program do the deconvolution. If your density maps would
> still indicate problems, you always can try to manually
> deconvolute.
> 
> By the way, your formulation  density> sounds like some cheap trick, when in fact you try to get
> a model that more accurately reflects your observed diffraction
> pattern.
> 
> Best, Herman
> 
> 
> -Urspr�ngliche Nachricht- Von: CCP4 bulletin board
> [mailto:CCP4BB@JISCMAIL.AC.UK] Im Auftrag von Chris Fage Gesendet:
> Dienstag, 6. Mai 2014 19:03 An: CCP4BB@JISCMAIL.AC.UK Betreff:
> [ccp4bb] Refining Metal Ion Occupancy
> 
> Hi Everyone,
> 
> In my 2.5-angstrom structure, there is negative Fo-Fc density
> surrounding a metal ion after refining in Phenix. From anomalous
> diffraction I am certain of the metal's identity and position in
> each monomer. Also, the ion is appropriately coordinated by nearby
> side chains. Should I be refining the occupancy of the ion in
> attempt to "flatten" the negative density? I am considering soaking
> the metal ion into crystals or cocrystallizing and collecting
> additional datasets.
> 
> Thanks for your help!
> 
> Regards, Chris
> 

- -- 
- --
Dr Tim Gruene
Institut fuer anorganische Chemie
Tammannstr. 4
D-37077 Goettingen

GPG Key ID = A46BEE1A

-BEGIN PGP SIGNATURE-
Version: GnuPG v1.4.12 (GNU/Linux)
Comment: Using GnuPG with Icedove - http://www.enigmail.net/

iD8DBQFTakicUxlJ7aRr7hoRAq9bAKCZURDNZxvfzuXpk0slF1vh6hkkLACfd5Ls
s7OV453Zj63k1xMOo+qzTfA=
=lpSS
-END PGP SIGNATURE-

Re: [ccp4bb] SOS from last century!

[Carter, Charlie]

Many thanks to Tim and Nat for, respectively, answering both questions fully, 
and to Ethan Merritt for subsequent help!!

Charlie

On May 6, 2014, at 1:24 PM, Tim Gruene wrote:

> Dear Charlie,
> 
> I still use molscript quite a bit, it is not so old fashioned as you
> might think. Raster3D is even in the standard repository of Debian.
> 
> You need to use the command line option '-r' in order to create Raster3D
> output. You can use the pipe
> 
> molscript -r -in your_molscipt.inp | render -png your_file.png
> 
> You get all command lines with 'molscript -h' and 'render -h'
> respectively, and their html documentation is really good if you need
> fine tuning.
> 
> Since png is state-of-the-art bitmap format, you can use the program of
> your choice for editing. I use gimp, if I need it at all, but sometimes
> also edit the Raster3D file manually, e.g. for setting labels.
> 
> Coot also creates output for Raster3D and also for Povray, in case this
> might be an option for you. That way you can easily include maps in the
> rendered file.
> 
> Regards,
> Tim
> 
> On 05/06/2014 07:15 PM, Carter, Charlie wrote:
>> I need help with a problem whose dimensions I perceive, but cannot surmount. 
>> 
>> It appears to be very important for me to re-make an illustration I made 
>> long ago using molscript. For various reasons, I cannot think of a way to do 
>> this with Pymol. I want to highlight active site residues by showing only 
>> the alpha carbons at approximately their van der Waals radii or a bit 
>> bigger. 
>> 
>> I resurrected molscript 2.1.2, recompiled it on my iMac and ran the input 
>> file, creating what appears to be a postscript file. Distiller converts it 
>> to a pdf file, but the image has all the wrong colors and hasn't been ray 
>> traced. The original illustration was prepared on a unix workstation that 
>> had a flow of programs that involved raster3D creating what I think were 
>> .png files, which I viewed and manipulated with a suite of unix-based public 
>> domain graphics utilities whose names I cannot recall, but they were much in 
>> vogue at the time. 
>> 
>> The header of the file output by molscript is:
>> 
>> %!PS-Adobe-3.0
>> %%BoundingBox: (atend)
>> %%Creator: MolScript v2.1.2, Copyright (C) 1997-1998 Per J. Kraulis
>> %%For: charleswcarterjr
>> %%DocumentNeededResources: font Times-Roman Symbol
>> %%Pages: 1
>> %%EndComments
>> %%BeginProlog
>> 50 dict begin
>> /R { setrgbcolor } bind def
>> 
>> which suggests it is a .ps file. 
>> 
>> Can anyone help me recover the rest of the software train that produced the 
>> images I once made? 
>> 
>> Or alternately how to create a similar view using pymol?
>> 
>> Many thanks,
>> 
>> Charlie
>> 
> 
> -- 
> Dr Tim Gruene
> Institut fuer anorganische Chemie
> Tammannstr. 4
> D-37077 Goettingen
> 
> GPG Key ID = A46BEE1A
> 

Re: [ccp4bb] AW: [ccp4bb] Refining Metal Ion Occupancy

[Bernhard Rupp]

>the negative difference density surrounding your metal ion shows that the 
>lower occupancy could not be fudged by a higher B-factor

Because there is a clear difference between high B-factor and low occupancy:
High B factor attenuates high resolution scattering most, while lower occupancy 
just evenly scales the scattering curve down.
Ergo, the FT - the Electron density - also looks different, with a low 
occupancy causing a WIDER scattering curve than a comparable high B-factor, 
thus transforming into a NARROWER peak compared to high B-factor. 

So, you could adjust (within physically meaningful limits) B and n to 'reshape' 
the electron density. If you have a negative difference density 'ring', your 
'observed' density there is less than the model density, 
and by reducing n you could reduce the wings of the model electron density 
peak, thus achieving a better match.

There is also the possibility that you have - perhaps in addition - some 
truncation ripples, which are most prominent around heavy atoms.

Figures 9-6 and 9-5 BMC.
This app allows to generate the different scattering curve shapes, and a 
similar app lets you FT it.
 http://www.ruppweb.org/new_comp/scattering_factors.htm


Best, BR

-Original Message-
From: CCP4 bulletin board [mailto:CCP4BB@JISCMAIL.AC.UK] On Behalf Of 
herman.schreu...@sanofi.com
Sent: Mittwoch, 7. Mai 2014 14:25
To: CCP4BB@JISCMAIL.AC.UK
Subject: [ccp4bb] AW: [ccp4bb] Refining Metal Ion Occupancy

Dear Chris,

In my experience, modern refinement program manage quite well to deconvolute 
occupancy and B-factor. In your case the negative difference density 
surrounding your metal ion shows that the lower occupancy could not be fudged 
by a higher B-factor. I would just refine occupancy and B-factor at the same 
time and let the refinement program do the deconvolution. If your density maps 
would still indicate problems, you always can try to manually deconvolute.

By the way, your formulation  sounds 
like some cheap trick, when in fact you try to get a model that more accurately 
reflects your observed diffraction pattern.

Best,
Herman


-Urspr�ngliche Nachricht-
Von: CCP4 bulletin board [mailto:CCP4BB@JISCMAIL.AC.UK] Im Auftrag von Chris 
Fage
Gesendet: Dienstag, 6. Mai 2014 19:03
An: CCP4BB@JISCMAIL.AC.UK
Betreff: [ccp4bb] Refining Metal Ion Occupancy

Hi Everyone,

In my 2.5-angstrom structure, there is negative Fo-Fc density surrounding a 
metal ion after refining in Phenix. From anomalous diffraction I am certain of 
the metal's identity and position in each monomer. Also, the ion is 
appropriately coordinated by nearby side chains. Should I be refining the 
occupancy of the ion in attempt to "flatten" the negative density? I am 
considering soaking the metal ion into crystals or cocrystallizing and 
collecting additional datasets.

Thanks for your help!

Regards,
Chris

Re: [ccp4bb] AW: [ccp4bb] Refining Metal Ion Occupancy

[Frank von Delft]

Don't forget, metals with their many electrons behave a lot better in 
refinement than light (CNOS) atoms, because scattering goes as square of 
number electrons.  So you ought to get good behaviour even for joint 
occupancy and B-factor refinement even at quite rubbish resolutions.



On 07/05/2014 13:25, herman.schreu...@sanofi.com wrote:

Dear Chris,

In my experience, modern refinement program manage quite well to deconvolute 
occupancy and B-factor. In your case the negative difference density 
surrounding your metal ion shows that the lower occupancy could not be fudged 
by a higher B-factor. I would just refine occupancy and B-factor at the same 
time and let the refinement program do the deconvolution. If your density maps 
would still indicate problems, you always can try to manually deconvolute.

By the way, your formulation  sounds 
like some cheap trick, when in fact you try to get a model that more accurately reflects your 
observed diffraction pattern.

Best,
Herman


-Ursprüngliche Nachricht-
Von: CCP4 bulletin board [mailto:CCP4BB@JISCMAIL.AC.UK] Im Auftrag von Chris 
Fage
Gesendet: Dienstag, 6. Mai 2014 19:03
An: CCP4BB@JISCMAIL.AC.UK
Betreff: [ccp4bb] Refining Metal Ion Occupancy

Hi Everyone,

In my 2.5-angstrom structure, there is negative Fo-Fc density surrounding a metal ion 
after refining in Phenix. From anomalous diffraction I am certain of the metal's identity 
and position in each monomer. Also, the ion is appropriately coordinated by nearby side 
chains. Should I be refining the occupancy of the ion in attempt to "flatten" 
the negative density? I am considering soaking the metal ion into crystals or 
cocrystallizing and collecting additional datasets.

Thanks for your help!

Regards,
Chris

[ccp4bb] AW: [ccp4bb] Refining Metal Ion Occupancy

[Herman . Schreuder]

Dear Chris,

In my experience, modern refinement program manage quite well to deconvolute 
occupancy and B-factor. In your case the negative difference density 
surrounding your metal ion shows that the lower occupancy could not be fudged 
by a higher B-factor. I would just refine occupancy and B-factor at the same 
time and let the refinement program do the deconvolution. If your density maps 
would still indicate problems, you always can try to manually deconvolute.

By the way, your formulation  sounds 
like some cheap trick, when in fact you try to get a model that more accurately 
reflects your observed diffraction pattern.

Best,
Herman


-Ursprüngliche Nachricht-
Von: CCP4 bulletin board [mailto:CCP4BB@JISCMAIL.AC.UK] Im Auftrag von Chris 
Fage
Gesendet: Dienstag, 6. Mai 2014 19:03
An: CCP4BB@JISCMAIL.AC.UK
Betreff: [ccp4bb] Refining Metal Ion Occupancy

Hi Everyone,

In my 2.5-angstrom structure, there is negative Fo-Fc density surrounding a 
metal ion after refining in Phenix. From anomalous diffraction I am certain of 
the metal's identity and position in each monomer. Also, the ion is 
appropriately coordinated by nearby side chains. Should I be refining the 
occupancy of the ion in attempt to "flatten" the negative density? I am 
considering soaking the metal ion into crystals or cocrystallizing and 
collecting additional datasets.

Thanks for your help!

Regards,
Chris

[ccp4bb] CCP4-6.4.0 Update 014

[charles . ballard]

Dear CCP4 Users

An update for the CCP4-6.4.0 series has just been released, consisting
of the following changes

* refmac5: 5.8.0071, fix for peptide planarity
* buccaneer: version 1.6 with new MR modes and user interface
* pointless: version 1.9.8, fixes for multilattice and output in qtrview
* aimiless: version 0.3.3, added option to analyse data against reference 
dataset
* ctruncate: version 1.15.4, ML target used in normalisation, bug fixes.
* viewhkl: Additional visual control to enhance weak reflections
* crank: fixes for mapro under os x/windows.

Note that auto-updates work only with CCP4 6.4.0 series, therefore
please upgrade if necessary.  The Update Manager is now included in the
package so you do not need to install it separately.  In addition, all
available updates will be installed automatically if you are using Setup
Manager for CCP4 installation.

Please report any bugs to c...@stfc.ac.uk.

Many thanks for using CCP4.

The CCP4 Core Team


-- 
Scanned by iCritical.

[ccp4bb] AW: SOS from last century!

[Herman . Schreuder]

It can be done with pymol:

In the box in the lower left corner is an item "selecting residues". By 
clicking on "residues" you can change it to "atoms" or "C-alphas"
When you now click on the appropriate active site atoms, you can select the 
alpha carbons and a "(sele)" item appears.
In the "(sele)" item you select "sphere" to get your spheres.
The radius of the spheres can be changed in "Setting" (pull-down menu at the 
top), "Edit All" and "sphere_scale". 

However, you may not get the nostalgic retro-look you might be after that way.

Herman



-Ursprüngliche Nachricht-
Von: CCP4 bulletin board [mailto:CCP4BB@JISCMAIL.AC.UK] Im Auftrag von Carter, 
Charlie
Gesendet: Dienstag, 6. Mai 2014 19:16
An: CCP4BB@JISCMAIL.AC.UK
Betreff: [ccp4bb] SOS from last century!

I need help with a problem whose dimensions I perceive, but cannot surmount. 

It appears to be very important for me to re-make an illustration I made long 
ago using molscript. For various reasons, I cannot think of a way to do this 
with Pymol. I want to highlight active site residues by showing only the alpha 
carbons at approximately their van der Waals radii or a bit bigger. 

I resurrected molscript 2.1.2, recompiled it on my iMac and ran the input file, 
creating what appears to be a postscript file. Distiller converts it to a pdf 
file, but the image has all the wrong colors and hasn't been ray traced. The 
original illustration was prepared on a unix workstation that had a flow of 
programs that involved raster3D creating what I think were .png files, which I 
viewed and manipulated with a suite of unix-based public domain graphics 
utilities whose names I cannot recall, but they were much in vogue at the time. 

The header of the file output by molscript is:

%!PS-Adobe-3.0
%%BoundingBox: (atend)
%%Creator: MolScript v2.1.2, Copyright (C) 1997-1998 Per J. Kraulis
%%For: charleswcarterjr
%%DocumentNeededResources: font Times-Roman Symbol
%%Pages: 1
%%EndComments
%%BeginProlog
50 dict begin
/R { setrgbcolor } bind def

which suggests it is a .ps file. 

Can anyone help me recover the rest of the software train that produced the 
images I once made? 

Or alternately how to create a similar view using pymol?

Many thanks,

Charlie

Re: [ccp4bb] Crystals Disappearing Overnight

[dusky dew]

I tried microbatch and the crystals are not stable. They dissolve
overnight.  I also have reproducibility issue.  Can this be due to poor
stability of adenosine?

Best
Maria

On Monday, May 5, 2014, Bob Cudney  wrote:
> Try the microbatch first to see if the problem is related to ionic
strength.
>
>
>
> When possible and practical it is good to change only one variable at a
time to identify cause and effect.
>
>
>
> Kind Regards, Bob Cudney
>
>
>
> Hampton Research
>
> 34 Journey
>
> Aliso Viejo, CA 92656-3317 USA
>
>
>
> Telephone 1 949 425 1321 Extension 200
>
> Fax 1 949 425 1611
>
> E-mail b...@hrmail.com
>
> Web www.hamptonresearch.com
>
>
>
> From: CCP4 bulletin board [mailto:CCP4BB@JISCMAIL.AC.UK] On Behalf Of
dusky dew
> Sent: Saturday, May 03, 2014 1:29 AM
> To: CCP4BB@JISCMAIL.AC.UK
> Subject: Re: [ccp4bb] Crystals Disappearing Overnight
>
>
>
> Thank you all for getting back!
>
> I will set up the drops using microbatch method.
> Regarding the temp, I set them up in lab and put them in incubator. The
lab temp may be slightly higher. So are they not stable at lower temp? Or
its the shock?
> So how can I take care of the temp issue?
>
> Thanks again!
> Maria
>
> On Saturday, May 3, 2014, Bob Cudney  wrote:
>> Hello Maria,
>>
>>
>>
>> Check to see if there might have been a temperature change between the
time the crystals were present and when the crystals disappeared.  If your
sample has temperature dependent solubility, in this relatively low ionic
strength condition, a temperature change could mean the difference between
the presence and absence of crystals.  That being said, if the experiment
is returned to the temperature that produced the crystals, the crystals
should/might reappear.
>>
>>
>>
>> If your drop is made by mixing 1 part of protein with 1 part of reagent
the initial drop concentration would be 10 mM Tris, 150 mM NaCl, 2.5% w/v
PEG 8,000, 50 mM Sodium cacodylate.  Is this a vapor diffusion experiment?
If yes, then the reservoir would be 5% w/v PEG 8,000, 100 mM Sodium
cacodylate.  The ionic strength of your drop would initially be higher than
the ionic strength in your reservoir.  This means water vapor leaves the
reservoir and vapor diffuses into the drop, lowering the protein and
reagent concentration in your drop.  This decrease in relative
supersaturation could dissolve a crystal.  Your set up would be a reserve
vapor diffusion.  You say the crystals appeared right after setting the
experiment so your crystallization is essentially a batch experiment.
Therefore you might want to change your set up from a vapor diffusion to a
microbatch experiment under oil.  If you need more information about how to
perform a microbatch experiment, let me know and I’ll explain.
>>
>>
>>
>> Hope this helps.
>>
>>
>>
>> Kind Regards, Bob Cudney
>>
>>
>>
>> Hampton Research
>>
>> 34 Journey
>>
>> Aliso Viejo, CA 92656-3317 USA
>>
>>
>>
>> Telephone 1 949 425 1321 Extension 200
>>
>> Fax 1 949 425 1611
>>
>> E-mail b...@hrmail.com
>>
>> Web www.hamptonresearch.com
>>
>>
>>
>> From: CCP4 bulletin board [mailto:CCP4BB@JISCMAIL.AC.UK] On Behalf Of
dusky dew
>> Sent: Friday, May 02, 2014 4:39 AM
>> To: CCP4BB@JISCMAIL.AC.UK
>> Subject: [ccp4bb] Crystals Disappearing Overnight
>>
>>
>>
>> Dear All,
>>
>>
>>
>> I am trying to crystallize a protein with Adenosine.  My protein is in
20 mM Tris, 300 mM NaCl and the crystals appear in a condition with 5
percent PEG8K, 0.1 M Sodium Cacodylate.  The protein is incubated with
adenosine for 1/2 hr before setting the drop.  The crystals appear right
after the drop is set but unfortunately they dissolve overnight.  The plate
is kept at 16 degree.
>>
>>
>>
>> Could anyone elaborate on this.  Is it possibly occurring because
Adenosine has stability issues.
>>
>>
>>
>> Thanks for your suggestions.
>>
>> ~ Maria
>>
>>
>>
>>

Re: [ccp4bb] SOS from last century!

[Karthikeyan S.]

Hi Charlie,

From molscript 2.1.2 you can directly create png, jpeg or gif also, 
default being postscript file. At least it is working in linux. I am not 
sure in iMac. Just type molscript -h. It should give the options...


HTH
-Karthik

On 5/6/2014 10:45 PM, Carter, Charlie wrote:

I need help with a problem whose dimensions I perceive, but cannot surmount.

It appears to be very important for me to re-make an illustration I made long 
ago using molscript. For various reasons, I cannot think of a way to do this 
with Pymol. I want to highlight active site residues by showing only the alpha 
carbons at approximately their van der Waals radii or a bit bigger.

I resurrected molscript 2.1.2, recompiled it on my iMac and ran the input file, 
creating what appears to be a postscript file. Distiller converts it to a pdf 
file, but the image has all the wrong colors and hasn't been ray traced. The 
original illustration was prepared on a unix workstation that had a flow of 
programs that involved raster3D creating what I think were .png files, which I 
viewed and manipulated with a suite of unix-based public domain graphics 
utilities whose names I cannot recall, but they were much in vogue at the time.

The header of the file output by molscript is:

%!PS-Adobe-3.0
%%BoundingBox: (atend)
%%Creator: MolScript v2.1.2, Copyright (C) 1997-1998 Per J. Kraulis
%%For: charleswcarterjr
%%DocumentNeededResources: font Times-Roman Symbol
%%Pages: 1
%%EndComments
%%BeginProlog
50 dict begin
/R { setrgbcolor } bind def

which suggests it is a .ps file.

Can anyone help me recover the rest of the software train that produced the 
images I once made?

Or alternately how to create a similar view using pymol?

Many thanks,

Charlie


--
*
Karthikeyan Subramanian, Ph.D.
Principal Scientist
CSIR-Institute of Microbial Technology
Council of Scientific and Industrial Research (CSIR)
Sector 39-A, Chandigarh - 160036, INDIA
email:skart...@imtech.res.in, skarthi...@gmail.com
Phone: +91-172-6665193 Fax: +91-172-2690585
*