Re: [ccp4bb]: Refmac vs. cns

[Jim Pflugrath <jwp@RigakuMSC.com>]

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

A historical perspective...
 Simulated annealling was quite the rage in the late 1980's.  At that
time, most electron density maps were fit by someone who had never fit an
electron density map (i.e. a student or post doc doing their very first
map).  The map-fitting programs had neither rotamer libraries, nor
fragments of main chain, nor any database automation that could be used as
a tool.  Thus, I believe many coordinates initially were fit rather badly
without regard to stereochemistry.  We needed a good refinement program
with a large radius of convergence to get atoms into the right position.
X-PLOR fulfilled that need.

Nowadays, with the tools provided by fitting programs and the knowledge of
our predecessors, fewer mistakes are made in the early interpretation
stages and those that are made are often discovered quickly.  Thus, there
is less need for simulated annealing to get out of false minimums.

Jim


>>Mark A. Wilson wrote (in part):
>...
> debate are of questionable validity. First, CNS (and earlier, XPLOR) were
> developed to address difficult refinement problems at low to moderate
> resolution (i.e. when data are very limited).  So, to claim that simulated
> annealing is best suited for higher resolution refinements is untrue and
> misses the declared purpose of the approach-to minimize the probability of
> a poor initial model being trapped in a local minimum when there are not
> enough data to guide the refinement to the global minimum.  This does not
> mean that you will never have to manually correct the resulting model, or
> to adjust the starting temperature for the refinement. It does mean that
> you don't have to rely solely on manual rebuilding to correct potentially
> major errors in an initial model that has been built into poor maps. The
>...