Hi All,
I suspect using a new free assignment for each data set is a fairly wide
spread mistake, which would be hard to pick up by a reviewer. It would
be interesting to survey the pdb for ligand bound structures which have
a different free R set from their parent (isomorphous) native
structures. I have heard from a colleague involved in testing
refinement programs that there is a surprising number of deposited
ligand structures where re-refinement shows that the ligand is built
into noise.
Would it be fair to say that in the case of a ligand structure,
isomorphous with the native structure but where different R-free sets
have been used, we cannot use the R-free to validate refinement of the
ligand? My understanding is that it would be contaminated by the phases
brought over with the model from the native data set, and is no longer
independent, so any ligand density could just arise from over fitting.
How could we test/retrieve such a situation? Will omit maps made after
a round of REFMAC refinement without the ligand still have a memory of
our over fitted model which would persist through to new maps? If so,
would this be the case after a slow cool in CNS?
Does the community still think it is valid to use a slow cool to "reset"
the free R?
Alun.
--
Alun R. Coker
University College London
Division of Medicine, Royal Free Campus
Centre for Amyloidosis and Acute Phase Proteins
Rowland Hill Street
London
NW32PF
Tel: +44(0)20 7433 2764
Fax: +44(0)20 7433 2776
