Precision and Reliability in Molecular
Structure Determination
J.Hašek, Institute of Macromolecular
Chemistry, Academy of Sciences of CR,
Heyrovského nám.2, 162 06 Praha 6, Czech
Republic
Protein
structure database (PDB) /1/ is a primary source of information
about the structure of biological macromolecules. It contains almost 20.000 of
experimentally determined structures. About 15 % of them are determined by NMR
techniques, 85 % by X-ray crystallography (see Tab.1).
Table 1. Number of structures
of biological macromolecules deposited in the Protein structure database. The theoretical models are not
collected in the PDB since 2002 (
information taken from the PDB Holdings
List: 04-Feb-2003)
|
Proteins Viruses |
Protein/NA Complexes |
Nucleic Acids |
Carbohydrates |
Details of measurement |
total |
X-ray |
15.507 |
734 |
638 |
14 |
8.755(i) |
16.893 |
NMR |
2.481 |
89 |
496 |
4 |
1.457(ii) |
3.070 |
Total |
17.988 |
823 |
1.134 |
18 |
10.212 |
19.963 |
(i)
Deposition
of reflection intensities
(ii)
Deposition of restrains gained by NMR measurement
In
spite of the fact that the final calculations and refinement of the structure with
NMR and X-ray data can be performed with the same computer program (e.g. XPLOR,
CNS /2/) there are significant differences in meaning and also in presentation
of structure.
NMR measurement. Roughly speaking, the most important information from
NMR measurement is identification of atoms laying near (3-6 Å) each other
in spite of the fact that they are far along the chain. The final description
structure is obtained by searching for all molecular models satisfying these
experimental restrains using the methods of molecular modeling. Thus, generally
speaking, the reliability of the structure model is derived from a completeness
of the experimental restrains, and the exact atom coordinates are optimized by methods
of molecular modelling. The molecular structure in PDB is described as a number
of individual structures often
interpreted as snapshots of a molecule in movement.
X-ray diffraction experiment. The primary
result of X-ray diffraction experiment is a map of electron density averaged
over time of measurement and all structure units in crystal. However, it is
really never published in this form. The atomic coordinates send to the PDB are
determined as centers of electron density
of individual atoms. Moving parts of molecule correspond to areas with low or
smashed electron density. At this moment, the X-ray scientist starts to look
for several alternative conformations which all are refined under the restrain
that the sum of occupation factors is 1. Thus the information about molecular
movement is hidden in a single file of atom coordinates as alternative
conformations for individual side chains and also as temperature factors B /3/ describing
the mean atomic displacement u [Ǻ] around the mean positions of individual
atoms. The dependence of B on the mean atomic displacement u [Ǻ]
is illustrated in Tab.2.
Tab. 2. Exact relation between the temperature factor
B [Å2] and the effective atom width (the mean atomic
displacement u [Ǻ]). B = 8 p2 <u2> .
B [Å2] |
4 |
8 |
16 |
32 |
64 |
128 |
Mean atomic displacement
[Å] |
0.23 |
0.32 |
0.45 |
0.64 |
0.90 |
1.80 |
Another term sometimes misunderstood
is resolution. The precision of atomic positions is not a simply function of
resolution and depends on more factors. An approximate relation between the
expected standard deviation of atomic position and the resolution is illustrated
in Tab.3.
Tab. 3. Typical
average coordinate inaccuracy <σx> (mean
expected standard deviation) as a function of the limit for
diffraction measurement (resolution). Data collected from randomly selected
structures found in literature.
Resolution [Å] |
5 |
3.0 |
2.4 |
1.9 |
1.5 |
1.3 |
1.0 |
0.8 |
0.6 |
Expected e.s.d.
[Å] |
> 3 |
0.7 |
0.4 |
0.2 |
0.1 |
0.07 |
0.05 |
0.03 |
0.01 |
The talk will show mutual
complementarity of X-ray and NMR
techniques and some rules for working with data obtained by X-ray
crystallography.
/1/ Protein DataBank (PDB). Research Collaboratory for Structural
Bioinformatics (RCSB) - .
http://rutgers.rcsb.org/pdb/
/2/ Giacovazzo C., Monaco
H.L., Viterbo D., Scordari F., Gilli G., Zanotti G., Catti M.
Fundamentals of
Crystallography. Oxford University Press, 2000.
/3/ International Tables for
Crystallography. Crystallography of
Biological Mactromolecules. Vol.F. Kluwer Acad.Publ.,Dorndrecht 1999.