As part of an INTAS project(04-77-7178) we have been looking at the
problem of protein structure matching. Proteins consist of sequences of
aminoacids ('sites') which form into complex three-dimensional
structures.
Both the chemical properties of the sites and the 3D structure are
crucial to the function of the protein in
biological systems. Proteins can have very different sequences but
perform similar functions by virtue of their
shape.
The goal of this work is to take a graph-matching approach to the
problem of comparing
these structures. The proteins are represented by graphs and are
matching using geometric and
relational information from the graph representation. Ultimately, we
hope to outperform systems
such as DALI and MATRAS in identifying protein similarity.