198 Topics in Current Chemistry Editorial Board: A. de Meijere KN ...

6 J.P. Glusker
2.1
Sources of Crystallographic Data
Two major requirements for meaningful comparisons of three-dimensional
structures for statistical analysis are:
1. ready access to the three-dimensional coordinates of all the appropriate
crystal structure determinations that contain the features chosen for comparison,
and
2. effective and easy-to-use analysis techniques.
Fortunately, efficient computer-based crystallographic databases containing
three-dimensional coordinates of atoms in crystal structures reported in the
scientific literature are now available to the scientific community. Therefore it is
no longer necessary to type large sets of numbers into a computer, because they
can be accessed in the correct format from the databases. In addition, there are
many excellent computer-graphics, geometrical, and statistical programs available
for application to these three-dimensional coordinates obtained from the
databases. This ensures that it is possible to compare structures. It is prudent,
however, for the investigator to build molecular models, as required, of the
ball-and-stick or space-filling variety, in order to obtain chemical or biochemical
insight from any comparisons that have been made.
The Cambridge Structural Database (CSD) [4] contains unit-cell dimensions
information on approximately 170,000 three-dimensional crystal structure
determinations that have been studied by X-ray or neutron diffraction. Each
crystal structure is identified by a unique six-letter code, called its REFCODE.
Duplicate structures and remeasurements of the same crystal structure are
identified by an additional two digits after the REFCODE. The CSD may be searched
in several ways. It is possible to find a list of bibliographic references to
reported data on all compounds with given chemical characteristics such as
steroids or peptides. In particular, however, the computer software provided
with the CSD allows one to search for a small group of atoms (either a full molecule
or a fragment of a chemical structure) with bonding that is precisely defined
by the codes that are provided by the user as input to the search. When
groups of atoms that meet the required specifications have been extracted from
the CSD, tables of required geometrical data may be generated and statistical
methods applied to the results. In analyses of hydrogen bonding the information
on crystal structures obtained by neutron diffraction (in which hydrogen atoms
are located with more precision than is possible for X-ray diffraction) are
important.
The Protein Data Bank (PDB) [5] is a computerized archive for the threedimensional
structural data on biological macromolecules – proteins and nucleic
acids. Each protein structure reported has an identifying code (IDCODE), a
header record containing useful information on the protein such as the name
and source of the protein and the resolution of the structure, together with a
series of references to published articles on the protein. Data are included on the
refinement methods used, such as the programs used, the R value, the number
of Bragg reflections, the root-mean-square deviations of the bond lengths and