Information Technology Reference
In-Depth Information
7. Conclusion
The thermodynamics and kinetics of hydrogen bond formation collude with the steric
constraints of the peptide bond to favour the formation of
-strands and reverse
turns along with a few less common secondary structures. Secondary structures are
characterized by the dihedral angles of the chain and the backbone hydrogen-bonding
pattern. The assignment of structure to a section of the chain by a pattern recognition
algorithm depends upon the definition of the hydrogen bonds and the boundaries of
φ
/
ϕ
space assigned for each structure.
The classification of folds in the PDB is achieved by segregating mainly-
α
-helices,
β
α
, mainly-
β
) structures. Structures in each class are then clustered by overall
shape and topology. Classification depends upon an overlap between structures in the same
group and a clear distinction between groups. However, while the mainly
and mixed (
α
/
β
and
α
+
β
α
-structures tend
to cluster separately, the highly populated architectures of the
-sheet- containing classes
tend to adopt similar two or three-layered sandwich-like structures, or barrels. Furthermore,
at the topology level the recurrence of certain structural motifs causes a significant overlap
between some folds. For example the repeated
βαβ
motif is found in both TIM barrels and
the doubly wound fold while the Greek-key is embedded in the jellyroll and the
immunoglobulin fold. Thus in some parts of fold space there is a continuum between
structures rather than distinct steps and in these regions, the criteria used for clustering
depends upon the purpose of the analysis.
CATH, SCOP and FSSP represent three unique ways of classifying protein
structure, FSSP uses a completely automated process, SCOP is principally derived from
manual inspection and CATH uses automated and manual procedures. Moreover whereas
FSSP and SCOP were created with an eye to evolutionary and functional relationships,
CATH was based solely on structural comparisons.
Another system, developed by Harrison et al., [100], classifies folds by their
'gregariousness' which is a measure of how many other folds have significant structural
overlap with a particular fold but have a different overall topology. In the analysis, folds in
the highly populated architectures, including the 10 so called super-folds, are highly
gregarious whereas folds such as
β
-helices comprising common motifs that are packed in
unusual ways, or folds with uncommon motifs, have low gregariousness. This method is
implemented by a graph-theoretic program, GRATH [100] that rapidly and accurately,
matches a novel structure against a library of domain structures to find the most similar
ones. It can be accessed via a server at http://www.biochem.ucl.ac.uk/cgi-bin/cath/Grath.pl.
GRATH is relatively fast and provides a reliable front-end filter for the more accurate, but
computationally expensive, residue based structure comparison algorithm SSAP, currently
used to classify domain structures in the CATH database.
β
Acknowledgements
AM is the recipient of an MRC Studentship. This work was supported, in part, by grant
B02959 from the BBSRC to BAW.
References
[1]
Aurora
,
R., Creamer, T.P., Srinivasan, R. & Rose, G.D., (1997) Local Interactions in Protein folding:
Lessons From The
α
-Helix.
J. Biol. Chem.
272, 1412-1416
