The biological question at hand determines which interpretation of the
X-ray data is needed. When studying protein mobility, one might prefer
complete structures, even if no evidence exists in the data to place
some residues anywhere. For dynamics calculations, antigenicity
prediction, etc, one might also prefer all B-factors to be realistically
related to the disorder/mobility rather than being residual B-factors
(over TLS). For drug design or homology modelling one might want to
resort to the most accurate coordinates.
So, there are many reasons why one might want to represent the X-ray
data differently from what the PDB provides. That is why we are making
our own PDBs. For example, in the BDB clone of the PDB all B-factors are
(as much as possible) full (isotropic) B-factors, while in the PDB_REDO
clone of the PDB most coordinates are more accurately in agreement with
the X-ray data. I think the PDB should stay as it is, and if questions
arise that require a different representation, just talk with us, and we
might even make it.
BDB: http://www.cmbi.umcn.nl/bdb/
PDB_REDO: http://www.cmbi.ru.nl/pdb_redo/index.html and
http://xtal.nki.nl/PDB_REDO/
Greetings
Gert
|