structural refinement using low temperature data

[Gerard 'CD' Kleywegt]

low temp:
- we've had good experiences with early refinement of an ovrall B
  in particular when starting from romm temp models or models
  with all Bs=20; in once case where Rfree appeared stuck at
  about ~42% and sim ann couldn't fix it, refinement of an
  overall B shift (grouped b refinement with group = (not hydrogen))
  reduced rfree by 4 or 5 % units and after that refinement proceded
  smoothly; reason of course: aver B is ~8 A2 for cryo model,
  but it was ~15 A2 for the starting model
- since at cryo temp many more waters are ordered you may need
  to put in a lot of them (if you have sufficient resolution)
  before rfree gets down to 'normal' levels
- we'll have to have a set of atomic resolution structures
  refined against cryo data to see if our models need to
  be adjusted
--gerard