Protein crystallography is an important source of data on 3-dimensional
structures of macromolecular complexes, which often are the biological
units performing a particular physiological function. Up to date, more
than 80% of PDB structures have been obtained by means of X-ray diffraction
on macromolecular crystals, yet the identification of complex structures is
problematic due to two main reasons. Firstly, only covalently linked
structures without crystallographic symmetry are identified reliably in
X-ray experiments, while monomeric units of macromolecular complexes are
linked by weaker, non-covalent interactions and are often crystallographi-
cally related. Secondly, it is generally unclear to what extent crystal
packing reflects structure of a complex as it is found in native environment.

In the talk, I will address these issues from the perspective of chemical
thermodynamics. A new approach to the identification of macromolecular
assemblies in crystal packings, which reaches 90% of correct identifications,
will be presented. I will also present results of docking studies on the
scale of the whole PDB, where the rate of failure may be interpreted in
thermodynamic context to indicate conditions under which complex structure
is reflected by crystal packing. The corresponding software is available as a
public web-server PISA at http://www.ebi.ac.uk/msd-srv/prot_int/pistart.html
and is deployed by the PDB as a mandatory processing tool since 2007.