Phasing power of cenrtic vs acentric reflections

Eleanor Dodson ccp4 at yorvic.york.ac.uk
Fri Feb 2 07:13:17 EST 1996


In article <1FEB96.17403227 at skyfox.usask.ca> matte at skyfox.usask.ca writes:

  >From: matte at skyfox.usask.ca
  >Newsgroups: bionet.xtallography
  >Date: 1 FEB 96 17:40:32 GMT
  >Organization: University of Saskatchewan
  >
  >Hello,
  >	In MIRAS calculations with MLPHARE from the CCP4 suite, I've
  >noticed that the phasing power (FH/<E>) for acentric reflections
  >is always greater than for the centric reflections for a given derivative. 
  >Its not intuitively obvious to me why this should be the case, as either 
  >the r.m.s. isomorphous difference would have to be lower, or the closure 
  >error larger, on average, for the centric than for acentric reflections.
  >This seems to make no sense, since the phase of the centric reflections
  >should in principle be uniquely pre-determined, and so the closure error
  >very small. This is certainly seen in the higher figures-of-merit
  >for centric vs acentric reflections, as one would expect. A recent paper 
  >by J.P. Abrahams and A.G.W. Leslie (Acta Cryst. D52, 30-42, 1996) on the 
  >structure determination of F1 ATPase indicated similar results with MLPHARE,
  >also in P212121 (compare PhPa and PhPc of Table 1). I must be missing 
  >something. Please enlighten me if you have some ideas about this.
  >
  >With thanks,
  >
  >Allan Matte
  >Dept.of Biochemistry
  >University of Saskatchewan
  >CANADA
  >
  >matte at sask.usask.ca 



     Hello.

Actually MLPHARE does not quote (FH/<E>) as the phasing power.  It
give (FH/(lack of closure)) in a resolution range. <FH> is much the
same for centric and acentric reflections, but <lack of closure> for
centrics should be proportional to <FH> + errors, while for acentrics
it should be proportional to <FH>/sqrt(2) + errors - that is related 
to the fact that <Diso> = <FH cos(random angle)>.. If your data was
error
free then you would see an exact sqrt ratio between the two - in fact
it isnt so the ratio will in general be < sqrt(2). 
  Eleanor Dodson





More information about the Xtal-log mailing list