[buster-discuss] Several questions - BUSTER-TNT refinement

Clemens Vonrhein vonrhein at globalphasing.com
Thu Aug 1 12:39:13 CEST 2013

Dear Jose,

good to hear from you - I hope everything is well in Oeiras. And:
sorry for the delay!

On Fri, Jul 19, 2013 at 02:01:12PM +0100, Jose Artur Brito wrote:
> I have a few questions regarding BUSTER-TNT refinement. I thought two times
> before sending them to the buster-discuss forum because they are probably
> too basic but in the end I decided to ask your help.

Good idea - let's hope we can give you that help ;-)

> So, here's the thing:
> - I solved the structure by SAD with the Fe anomalous signal with data
> collected at 1.72A wavelength (the structure is to 1.98A and also has
> two iodide atoms which were useful in phasing as extra sites);
> during the refinement, I saw that there are two keywords in BUSTER that
> I thought it could useful and used them: "wavelength=1.72" and
> "AtomicFormFactorCorrection=yes"; was this OK, did I do the correct thing?


With two points to be aware of:

  a) usually your input MTZ file should already contain the correct
     wavelength in the header (since that information is present
     during data-processing). If it doesn't (but contains something
     funny like 0.000, 1.000 or CuKa): it might be worthwhile checking
     your data-processing pipeline to see where it 'looses' that

     Needless to say that autoPROC would create MTZ files with the
     correct wavelength value ;-)

  b) Since you collected close to the Fe edge, the values BUSTER would
     use to automatically correct your atomic formfactors, will come
     from a simple theoretical calculation using CROSSEC. This can be
     quite a bit off, if your actual fluorescence scan looks different
     (which it does most of the time). So you could also set the
     values explicitely (from a fluorescenec scan) via the
     FormfactorCorrection parameter. However, that might be tricky
     when you have a mixture of I and Fe atoms (in the fluorescence

> - now, I'm refining a second structure of the same protein but this time
> reduced with dithionite; the resolution is 1.75A and the data was
> collected at 1.23A wavelength; should I also use those two keywords?
> This structure has the two Fe atoms and 1 iodide atom.

It should never hurt using that keyword
(AtomicFormFactorCorrection=yes): if there is negligible (theoretical)
f' value it will result in ... negligible correction. Just monitor the
messages you are getting from BUSTER anout what it actually did when
switching on that feature.

> - is there any detrimental effect in using an "nmiss" higher that what
> it is expected? I don't see ca. 20 residues at both termini and and a
> loop of about 6 residues; I used "nmiss=1000" in the initial rounds of
> refinement and reduced to "nmiss=300" in this last refinement;

Very good question: a very small missing part in general will/should
have only a small impact on the overall scaling anyway - and since it
is used as a uniform prior, it gets deluted further. It will help
scaling a bit through - but using nres*11 or nres*11*2 (nres = number
of residues with an average of 11 atoms) shouldn't make a big
differences when it is a small part of the overall model.

> - should I provide a sequence file to the program in order to tell it
> which residues are actually missing? What is the keyword for this?

No: that doesn't make a difference: one can model everything as
oxygens and it is a uniform prior, not something that is localised in
a particular, known region.

> - the structure is a cytochrome and when refining the heme, BUSTER
> corrects the format of COOT with "pad with spaces" and correct to
> "nomenclature v2"; what is this? is there a conflict between the
> libraries of both programs?

Yes: the current BUSTER release uses internally still PDB v2 atom
nomenclature. The next release will be completely PDB v3 from input to

Does that help?


Clemens (for buster-develop)


* Clemens Vonrhein, Ph.D.     vonrhein AT GlobalPhasing DOT com
*  Global Phasing Ltd.
*  Sheraton House, Castle Park 
*  Cambridge CB3 0AX, UK
* BUSTER Development Group      (http://www.globalphasing.com)

More information about the buster-discuss mailing list