your questions are well-justified and I'm looking forward to the
comments of others.
My own impression when looking at the B-values of your fourth round PDB
file is that they appear quite reasonable for the protein: smooth along
the main chain and increasing sharply along the side chains. For those
few cases where B rises to a maximum and then drops again when "looking
along the side chain" there might be a problem with the conformation of
the side chain, and/or alternative conformations.
The problem with B factor refinement, or even more, with the way it is
currently done, is that the rmsd values refer to absolute differences,
whereas they should refer to relative differences of B factors. In other
words, a residue with B-values around 60 should be able to adopt rms
differences between Bs of bonded atoms around maybe 6, in the same way
as a residue with Bs around 20 should be able to display rmsds around 2.
This is, however, not the way B refinement is currently programmed.
I do think that B values around 80 of waters might apply for mobile
waters; I would not agree to a 50 or 60 A**2 cutoff. Also the sulfate Bs
appear reasonable (didn't look at your coords on the graphics though);
if O4 is the attachment to the protein then O1-O3 might well have the
So, at the resolution of your data and provided the data are good, those
B values are probably ok, as indicated by R_free!
email: Kay.Diederichs at uni-konstanz.de Tel +49 7531 88 4049 Fax 3183
Fakultaet fuer Biologie, Universitaet Konstanz
Postfach M656, D-78457 Konstanz, Germany