[phenixbb] calculating anomalous differences from a model

Randy Read rjr27 at cam.ac.uk
Fri Apr 4 03:45:05 PDT 2014


Hi,

The syntax and behaviour have changed since 2008.  Most of the time, people run LLG completion until Phaser finds no more sites, in which case the log-likelihood-gradient map is by definition close to flat at that point.  So the default is not to write out the map coefficients.  A special flag (LLGMAPS ON) has to be set to get the map coefficients, and then they go into a separate file.

Here's a sample script to make an initial log-likelihood-gradient map for our MR-SAD tutorial:

phenix.phaser << eof > LLGmap.log
ROOT LLGmap
TITLE MR_SAD starting from MR sol with goat alpha-lactalbumin model
MODE EP_AUTO
HKLIN lyso2001_scala1.mtz
LLGCOMPLETE COMPLETE ON
LLGCOMPLETE NCYC 0
LLGMAPS ON
LLGCOMPLETE SCATTERER S
PARTIAL PDB eptute_1.1.pdb IDENT 0.4
CRYSTAL four DATASET CuKa LABIN F+ = F_CuKa(+) SIG+ = SIGF_CuKa(+) F- = F_CuKa(-) SIG- = SIGF_CuKa(-)
COMPOSITION PROTEIN SEQ hewl.pir NUMBER 1
WAVELENGTH 1.5418
eof

You have to say "LLGCOMPLETE COMPLETE ON" to get the LLG map calculated, but "LLGCOMPLETE NCYC 0" to avoid any sites being added, and "LLGMAPS ON" to get the output LLG map, which has coefficients FLLG_S and PHLLG_S for this case (with "S" for the type of scatterer).

However, we've tested various options as part of the work on ion identification (led by Nat Echols), and if you want to identify anomalous scatterers from a final model it's better to do iterative completion adding purely imaginary scatterers to a model of real scatterers.  The refined occupancy gives you a measure of the f" (or f" times occupancy for a partially occupied anomalous scatterer) for the site identified by the imaginary scatterer location, and by doing iterative completion the inclusion of sites found in the initial cycles improves the signal-to-noise in later cycles.  Here's a script that could be used for a final model:

phenix.phaser << eof > MRSAD_imag.log
ROOT MRSAD_imag
TITLE MR_SAD adding purely imaginary scatterers to final model
MODE EP_AUTO
HKLIN lyso2001_scala1.mtz
LLGCOMPLETE COMPLETE ON
LLGCOMPLETE SCATTERER AX
PARTIAL PDB HEWL.pdb IDENT 1.0
CRYSTAL four DATASET CuKa LABIN F+ = F_CuKa(+) SIG+ = SIGF_CuKa(+) F- = F_CuKa(-) SIG- = SIGF_CuKa(-)
COMPOSITION PROTEIN SEQ hewl.pir NUMBER 1
WAVELENGTH 1.5418
eof

Here what you want to look at is the set of sites (MRSAD_imag.1.pdb) with their occupancies, and not an LLG map (which isn't produced by this script).  There are occasionally some spurious sites, especially if the anomalous signal is very weak, but it usually provides good evidence when a site falls on top of what you expect to be a metal ion.

Best wishes,

Randy

On 3 Apr 2014, at 17:28, Alastair Fyfe <afyfe at SOE.UCSC.EDU> wrote:

> A related question: is there a current example of  using phaser to calculate SAD LLG maps from a refined model not refined with phenix.refine? In the past I have used the description posted here:
> http://www.phenix-online.org/pipermail/phenixbb/2008-July/012399.html
> 
> However phaser syntax and keywords seem to have changed a bit as this no longer works.  Also, have the derived FLLG and PHLLG columns been renamed ?
> thanks,
> Alastair Fyfe
> 
> On 04/03/2014 09:13 AM, Nathaniel Echols wrote:
>> PS. Tom pointed out that the anomalous measurability in Xtriage depends on
>> the sigmas, which is obviously a problem for synthetic data - you can
>> generate fake sigmas with "add_sigmas=True", but the resulting statistics
>> will be meaningless.
>> 
>> -Nat
>> 
>> 
>> On Thu, Apr 3, 2014 at 8:27 AM, Nathaniel Echols <nechols at lbl.gov> wrote:
>> 
>>> I guess it depends on what you're looking for as the final output.  It's
>>> easy to generate an MTZ file with anomalous Fcalc (this is in the GUI too,
>>> of course):
>>> 
>>> phenix.fmodel model.pdb high_resolution=2.0 type=real wavelength=0.9792
>>> 
>>> Extracting some kind of useful summary from the data might require a
>>> little extra scripting - although this may be the kind of thing we should
>>> just add to Xtriage (which only reports "anomalous measurability" right
>>> now).
>>> 
>>> -Nat
>>> 
>>> 
>>> 
>>> On Thu, Apr 3, 2014 at 7:20 AM, Jonathan Grimes <jonathan at strubi.ox.ac.uk>wrote:
>>> 
>>>>    Given a refined protein structure, is there an straightforward way to
>>>> calculate the anomalous
>>>>    differences as a function of resolution, at wavelength X.
>>>> 
>>>>    many thanks
>>>>    jon
>>>> 
>>>> Dr. Jonathan M. Grimes,
>>>> NDM Senior Reseach Fellow
>>>> University Research Lecturer
>>>> DIAMOND Research Fellow
>>>> 
>>>> Division of Structural Biology
>>>> Wellcome Trust Centre for Human Genetics
>>>> University of Oxford
>>>> Roosevelt Drive,
>>>> Oxford OX3 7BN, UK
>>>> 
>>>> Email: Jonathan at strubi.ox.ac.uk, Web: www.strubi.ox.ac.uk
>>>> Tel: (+44) - 1865 - 287561, FAX: (+44) - 1865 - 287547
>>>> 
>>>> _______________________________________________
>>>> phenixbb mailing list
>>>> phenixbb at phenix-online.org
>>>> http://phenix-online.org/mailman/listinfo/phenixbb
>>>> 
>>> 
>> 
>> 
>> _______________________________________________
>> phenixbb mailing list
>> phenixbb at phenix-online.org
>> http://phenix-online.org/mailman/listinfo/phenixbb
> 
> _______________________________________________
> phenixbb mailing list
> phenixbb at phenix-online.org
> http://phenix-online.org/mailman/listinfo/phenixbb

------
Randy J. Read
Department of Haematology, University of Cambridge
Cambridge Institute for Medical Research      Tel: + 44 1223 336500
Wellcome Trust/MRC Building                   Fax: + 44 1223 336827
Hills Road                                    E-mail: rjr27 at cam.ac.uk
Cambridge CB2 0XY, U.K.                       www-structmed.cimr.cam.ac.uk

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://phenix-online.org/pipermail/phenixbb/attachments/20140404/e44b700f/attachment.htm>


More information about the phenixbb mailing list