I would like to point out that it might be best to refine the structure
without any electrostatic interactions (flags ... excl elec pele end).
This removes any possible bias imposed by the force field. The idea
is that if you have good enough diffraction data it should produce
the correct distances and chemical relationships between atoms.
Refinement without electrostatic term is of course used in PROLSQ,
TNT, and other programs. Thus, by turning off the electrostatics
in X-PLOR, you'll get very similar behavior of the programs (assuming
you're using the Engh and Huber force field).
Note, that electrostatics is not required for annealing. The protocols
in X-PLOR are completely independent of the force field (or parameter
dictionary). In fact, by turning off the electrostatics you might
get better convergence.
I would also recommend to try using the repel option throughout the
annealing stage. Our most recent experience indicates that
this helps in difficult cases even when used with normal Verlet (Cartesian)
dynamics (cf. Rice & Brunger, Proteins 19, 277--290, 1994).
PS: Please note that there is still a problem with the US bionet mail server.
I've now subscribed to the site in the UK. The UK site seems to work fine. However,
the US and the UK mail server don't seem to talk to each other.