Re: [ccp4bb] Why the phases from sftools are flipped?

[Bruno KLAHOLZ]

Dear Steven,

maybe you have to permutate the order in the way the map is read in, here is 
what I have used some time ago for an EM map.

HTH, best regards,

Bruno


/usr/local/rave/rave_linux/lx_mapman < set spacegroup
> P1
> set cell
> 200  200 200   90.00090.00090.000
> mapin input.map MAP
> map2sf resol  3  column F_prot PHI_prot
> write output.mtz
> exit

Was sftoold designed in this way?

Thanks!
Steven


--
Steven Chou

[ccp4bb] Why the phases from sftools are flipped?

[Steven Chou]

Dear All,

I used 'sftools (in CCP4)' to convert a map (in CCP4 format) to structure
factors (.mtz).
However the phases were flipped. I back converted it into a map file and
compared it with the original one. The back converted one was just the
mirror of the original one. To get the right phase, I had to convert the
structure factors to a text format (CIF or XPLOR) and edit it with script.
Their relationship is: CorrectPhase = 360 - phaseFromSftools.

I also used CNS to do the same conversion. The phases from CNS were correct.

sftools
> set spacegroup
> P1
> set cell
> 200  200 200   90.00090.00090.000
> mapin input.map MAP
> map2sf resol  3  column F_prot PHI_prot
> write output.mtz
> exit

Was sftoold designed in this way?

Thanks!
Steven


-- 
Steven Chou

[ccp4bb] How to create a mask with specified pixel size in CCP4?

[Steven Chou]

Dear All,
I'm trying to generate a mask from a pdb coordinate file using the
'Ncsmask' utility of CCP4 (Map & Mask Utilities => Create/Edit Masks).
The parameters I used are:
=
Radius for building mask from atoms 10.0 angstrom.
Space group P1
Set map extent x 0.0  307.2, y 0.0  307.2, z 0.0  307.2
Set map grid x=128 y=128 z=128

Cleanup mask to have 1 contiguous region
Trim maks to minimum box
=
I want to have a mask with pixel size of 307.2/128 =2.4, and box size of
128.

But the mask I got has a pixel size of 1 A/pixel, and box size of 308.

Is there anything wrong with my parameters?
It seems that the mask generated by Ncsmask always has a pixel size of 1.

Thanks in advance!
Steven

[ccp4bb] Complementary Non-diffraction Techniques in Structural Biology, 11th December 2014, Brunei Gallery , SOAS, Russell Square, London

[Antonyuk, Svetlana]

  2014 BBS Symposium

Complementary Non-diffraction Techniques in Structural Biology
A one-day symposium to celebrate the end of the
"UN International Year of crystallography 2014"

Date: 11th December 2014 - from 10.30 to 16.30
Venue: Brunei Gallery , SOAS, Russell Square, London WC1H 0XG

Anthony Watts - 
Chairman, British Biophysical Society
Speakers:

Olwyn Byron - University of 
Glasgow
Richard Henderson - MRC Laboratory of Molecular Biology, 
Cambridge
Steve Matthews - Imperial College, 
London
Sheena Radford - University of 
Leeds
Carol Robinson - University of 
Oxford
Mark Sansom - University of 
Oxford
John Walker - University of 
Cambridge
Bonnie Wallace, Birkbeck College 

To register for the symposium please choose one of the options below:

  *   BBS Members, £10 


  *   Non-members of the BBS, 
£40

  *   Speaker / Honorary  member, no 
charge

payable in advance before 21 November 2014.

This fee includes a hot buffet lunch and refreshments for all registered 
participants.

To become a member of the BBS (£25 annual membership), please complete the 
applications forms found 
here.

--

Dr. Svetlana Antonyuk
Senior Lecturer
Institute of Integrative Biology
University of Liverpool
Biosciences Building
Liverpool
L69 7ZB
--
Phone:   +44(0)151-795 0349
Email:s.anton...@liverpool.ac.uk

Re: [ccp4bb] Refinement of Iron-sulphur clusters

[George M. Sheldrick]

Dear Oliver,

Since you mentioned my name let me try to confuse the issue. The iron
atoms in Fe4S4 clusters can adopt different oxidation states, e.g. in
HiPIPs and Ferrodoxins, and one might expect this to influence the
geometry of the clusters. So maybe you would need several different sets
of restraints. However at the time most of these structures were
determined, the influence of radiation damage was generally
underestimated. In general the first thing that happens in a synchrotron
beam is that the clusters mop up the electrons released by the action of
the radiation and so they probably had lower oxidation states than the
people determining the structures (at least this one) thought that they
had. What we should have done was to try to get a Moessbauer spectrum of
each crystal before and after the data collection, but it is easy to say
that now and anyway the crystals were probably too small.

Best wishes, George

> 
> The dictionaries I sent have values that are obtained from an analysis of
> small
> molecule crystal structures in the CSD (by "hand" with conquest). For
> instance
> the Fe-Scluster-Fe angle has a mean of  73.4 degree with a standard
> deviation
> of 0.7 degrees. The restraints are compatible with Engh and Huber restraints
> you
> will be using for your protein and for that matter the Parkinson et al
> restraints used
> for any nucleic acid (because they come from a similar data mining
> procedure).
> So if Stephen used this dictionary he would not " introduce a bias in the
> final rms
> angle deviation value" because the restraint values would be sensible.
> 
> If you to check the restraint values are sensible then you check them
> against atomic
> resolution structures for instance
> http://www.rcsb.org/pdb/explore/explore.do?structureId=1B0Y
> at 0.93 Å resolution structure determined by George Sheldrick. This has SF4
> with
> Fe-Scluster-Fe angles around 73.3 degrees.
> 
> I would strongly advise anyone against using a restraint dictionary for
> Fe4S4 that uses
> 90 degree as the ideal angle because you are restraining towards values that
> 
> make no sense. If the dictionary contains chiral volume terms then I would
> ask the
> question why? But this is just my personal take.
> 
> If CCP4 would like to redistribute/work the dictionaries sent then we would
> be happy
> (please email off-list).
> 
> Dr Oliver Smart
> Director SmartSci Limited http://www.smartsci.uk/
> & Consultant Global Phasing Ltd http://www.globalphasing.com/
> 

-- 
Prof. George M. Sheldrick FRS
Dept. Structural Chemistry,
University of Goettingen,
Tammannstr. 4,
D37077 Goettingen, Germany
Tel. +49-551-39-33021 or -33068
Fax. +49-551-39-22582

Re: [ccp4bb] incorrect R-factor calculation in sftools

[Tim Gruene]

Hi Kay,

ok, fair point.

Best,
Tim

On 10/14/2014 01:02 PM, Kay Diederichs wrote:
> Tim,
> 
> I would not consider this as incorrect! What sftools calculates is R_scale, a 
> symmetric version of an R-value. Makes sense when comparing two data sets, or 
> two model amplitudes. Clearly, it gives values which differ from model R 
> values. But at least it defines what it prints out.
> 
> I put this into the wiki ( 
> http://strucbio.biologie.uni-konstanz.de/ccp4wiki/index.php/R-factors ).
> 
> best,
> 
> Kay
> 
> 
> On Tue, 14 Oct 2014 02:49:23 +0200, Tim Gruene  
> wrote:
> 
>> Dear all (dear developers),
>>
>> on a recent discussion on the phenixbb, Nat figured out that sftools
>> calculates the R-factor incorrectly as
>>
>> "200*Sum|col1-col2|/sum(col1+col2)"
>>
>> instead of
>>
>> "100*Sum|col1-col2|/sum(col1)"
>>
>> May I suggest to either correct this or not call it Rfactor in order to
>> avoid future confusion (as it did on the phenibb)?
>>
>> Best wishes,
>> Tim
>> -- 
>> Dr Tim Gruene
>> Institut fuer anorganische Chemie
>> Tammannstr. 4
>> D-37077 Goettingen
>>
>> GPG Key ID = A46BEE1A
>>
>>
> 

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

GPG Key ID = A46BEE1A



signature.asc
Description: OpenPGP digital signature

Re: [ccp4bb] Refinement of Iron-sulphur clusters

[Oliver Smart]

on 13/10/14 6:33 PM, Pedro Matias  wrote:

> In relation to this topic, I'd like to mention that SF4 has replaced FS4
> as the Fe4S4 monomer in the CCP4 monomer library.
>
> However, the dictionary values for bond lengths and angles are not
> correct, and this is especially noticeable when the dictionary is used
> in a high-resolution refinement (e.g., 1.05 A). In particular, the
> Fe-S-Fe angles are all set to 90 degrees, when in fact some are smaller
> and others are larger.
>
> This introduces a bias in the final rms angle deviation values. Fe4S4 is
> a distorted cubane, not a perfect cube.
>

Pedro,

Yes you are 100% correct

The dictionaries I sent have values that are obtained from an analysis of
small
molecule crystal structures in the CSD (by "hand" with conquest). For
instance
the Fe-Scluster-Fe angle has a mean of  73.4 degree with a standard
deviation
of 0.7 degrees. The restraints are compatible with Engh and Huber restraints
you
will be using for your protein and for that matter the Parkinson et al
restraints used
for any nucleic acid (because they come from a similar data mining
procedure).
So if Stephen used this dictionary he would not " introduce a bias in the
final rms
angle deviation value" because the restraint values would be sensible.

If you to check the restraint values are sensible then you check them
against atomic
resolution structures for instance
http://www.rcsb.org/pdb/explore/explore.do?structureId=1B0Y
at 0.93 Å resolution structure determined by George Sheldrick. This has SF4
with
Fe-Scluster-Fe angles around 73.3 degrees.

I would strongly advise anyone against using a restraint dictionary for
Fe4S4 that uses
90 degree as the ideal angle because you are restraining towards values that

make no sense. If the dictionary contains chiral volume terms then I would
ask the
question why? But this is just my personal take.

If CCP4 would like to redistribute/work the dictionaries sent then we would
be happy
(please email off-list).

Dr Oliver Smart
Director SmartSci Limited http://www.smartsci.uk/
& Consultant Global Phasing Ltd http://www.globalphasing.com/

Re: [ccp4bb] coot: protein sequence number too large to be displayed

[Paul Emsley]

There is not a straightforward fix. There is a limit of 2^15 on pixel 
indexing on (some?) X11 servers, I believe. If you compile your own 
coot, you can try changing scroll_width_max in src/nsv.cc to 64700.


Paul.

On 10/10/14 14:52, Yong Tang wrote:
Thanks Tim for the input - so far you are the only person to response 
so I assume there is not a more straightforward solution to it - like 
changing a parameter in .coot to allow display of a larger number of 
residues. Changing and re-changing the residue numbers should work 
fine for an isolated structure, but may be prone to human mistakes 
when this process is required for each every structure in a 
high-throughput setting. Sincerely yours, -yong


On Thu, Oct 9, 2014 at 5:06 PM, Tim Gruene > wrote:


Dear Yong,

one work-around is to renumber the residues, e.g. with Coot itself or
with pdbset. This, however, would break the reasonable custom that
residue number should match the biological sequence number, so don't
forget to rerenumber upon deposition.

Regards,
Tim

On 10/09/2014 08:36 PM, Yong Tang wrote:
> Hi there I have a quick question about coot. I'm using Coot
0.8-pre EL
> (revision 5001). I just realize that the sequence display is
limited to
> somewhere around 2245... The structure that I am looking at has the
> starting residue number larger that 2245 so the whole sequence
is not shown
> in Draw/Sequence View... Any help to get around is greatly
appreciated.
> Thank you, -yong
>

Re: [ccp4bb] incorrect R-factor calculation in sftools

[Kay Diederichs]

Tim,

I would not consider this as incorrect! What sftools calculates is R_scale, a 
symmetric version of an R-value. Makes sense when comparing two data sets, or 
two model amplitudes. Clearly, it gives values which differ from model R 
values. But at least it defines what it prints out.

I put this into the wiki ( 
http://strucbio.biologie.uni-konstanz.de/ccp4wiki/index.php/R-factors ).

best,

Kay


On Tue, 14 Oct 2014 02:49:23 +0200, Tim Gruene  
wrote:

>Dear all (dear developers),
>
>on a recent discussion on the phenixbb, Nat figured out that sftools
>calculates the R-factor incorrectly as
>
>"200*Sum|col1-col2|/sum(col1+col2)"
>
>instead of
>
>"100*Sum|col1-col2|/sum(col1)"
>
>May I suggest to either correct this or not call it Rfactor in order to
>avoid future confusion (as it did on the phenibb)?
>
>Best wishes,
>Tim
>-- 
>Dr Tim Gruene
>Institut fuer anorganische Chemie
>Tammannstr. 4
>D-37077 Goettingen
>
>GPG Key ID = A46BEE1A
>
>