[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [ccp4bb]: REFMAC5 maximum likelihood refinement



***  For details on how to be removed from this list visit the  ***
***          CCP4 home page http://www.ccp4.ac.uk         ***

Dear Wolfram:

I am new to refmac so I can hardly be considered and authority.  But my 
understanding is the idea behind m.l. refinement is to do no more 
refinement than is justified, whereas minimizing a crystallographic 
residual (in the conventional manner) tends to over-refine the data.  
I've noticed both in CNS and in refmac that conventional 
crystallographic residual refinement always gives me lower R-factors 
than does m.l. refinement, all other things being equal.  This does not 
mean that the refinement is better.  The maps, and therefore the 
structure, will be less biased.  Also I wouldn't call a 2% increase 
"dramatic."  Differences in the methods for calculating a solvent mask, 
different libraries for scattering amplitudes, geometric parameters, 
etc. could also easily account for some if not all of the 2% 
difference.  FWIW I am becoming convinced that refmac works better than 
CNS for a final refinement, at least in my hands.

Bill Scott


On Wednesday, July 3, 2002, at 04:03 PM, Wolfram Tempel wrote:

> ***  For details on how to be removed from this list visit the  ***
> ***          CCP4 home page http://www.ccp4.ac.uk         ***
>
> Hi all,
> under which circumstances would maximum likelihood target refinement in
> REFMAC5 produce a dramatic INCREASE in R (from 0.187 to 0.208) and Rfree
> (from 0.232 to 0.257) with 1.8A data? This was not accompanied by
> improved geometry (reduced RMSs). Any idea where I should start looking
> for the underlying problem?
> Thank you.
> Wolfram Tempel
> U. of Georgia
>
>
>
>