[buster-discuss] why 02-BUSTER ?

Clemens Vonrhein vonrhein at globalphasing.com
Thu Jun 24 13:55:59 CEST 2010

Hi Jonathan,

On Thu, Jun 24, 2010 at 11:02:18AM +0100, Jonathan Grimes wrote:
>   Can someone explain why BUSTER does a 01-BUSTER
>   and then seems to redo refinement creating a 02-BUSTER
>   directory.  What is the decision making process ??

The 'refine' command loops over all F/SIGF pairs in your MTZ file (or
more correctly: all pairs of F/Q column type). The final refine.pdb
will be the one with the best Rfree.

If you want to restrict that:

  % refine autoBUSTER_fss="Fnew,SIGFnew"

which would set the F/SIGF pair to use explicitely (and ignore all

>    I have been running BUSTER on a protein with 3-fold   
>    ncs, so using the -autoncs flag and using grouped B-factor
>    refinement (the data go to 3.5A), with 4 big cycles and 50 small.
>    Previously BUSTER only ran creating 01-BUSTER.  Now
>    it appears to repeat the process, with the Rfactor jumping
>    back upto the value (close to, bit not exactly) at the start
>    of 01-BUSTER.

Yes, when it picks the second set of F/SIGF values, it goes back to
the same starting point. But since the reflections are actually
different, thet statistics will be different too.

The easiest is to always use the same 'master' reflection file for
refinement - ideally the one coming out of TRUNCATE (which you
probably did run at some stage) and after adding the FreeR_flag
column. Using some other file (output MTZ from another refinement,
after scaling to another dataset etc etc) is always dangerous:

 * it can mess up test-set flags

 * you could be loosing reflections

 * mess up the cell parameters in the MTZ file (the order of files
   e.g. given to CAD matters!)

 * introduce additional scaling

Always good to stay with the data as close to the original images as
possible - and those MTZ files will most likely contain only a single
F/SIGF pair.




* 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