*** For details on how to be removed from this list visit the ***
*** CCP4 home page http://www.ccp4.ac.uk ***
A (very) long shot might be pseudosymmetry - what were the Rmerge
statistics like? How did they compare with monoclinic? A dimer in a
monoclinic space group with non-crystallographic symmetry that closely
approximates orthorhombic crystallographic symmetry might exhibit this
behaviour.
I'm not sure it's that much of a long shot - I had almost this exact
problem recently, which caused much anguish and many similar emails to
ccp4bb. There was nothing in the processing statistics to suggest that
anything was wrong, and the maps looked great, but Rfree was considerably
higher than it should have been. This was in P212121, and although P21
did not scale appreciably better, it yielded a much better Rfree. My
own experience is that lower symmetry space groups will usually have a
better Rsym anyway, so you can't always rely on that. (You probably need
to try P21 with all three possible screw axes - I think only one will be
correct if you have psuedosymmetry.)
It might help to see a sample diffraction image or two; in my case the
problem was even weirder than the above simple explanation indicates, and
I was only able to figure it out when I looked more carefully at the spots
and realized that every other lattice line was very weak and noisy. I
still haven't figured out what kind of pseudosymmetry that was.
-Nat
(The only caution is that NCS restraints in these cases can break
cross-validation, so you need to pick the Rfree set carefully to
avoid that. I used SFTOOLS to pick thin resolution shells for Rfree, but
supposedly there's a way to pick NCS-related reflection pairs instead - or
something like that - but I'm not sure which program does this.)
