Project VIII: Predicting protein-protein interaction
using statistical classification methods
Virtually all cellular processes
depend on precisely organized interactions between proteins, and a critical
step in understanding the physiological function of a protein is the identification
other proteins with which it interacts. Characterization of protein interactions
is important for problems ranging from rational drug design to analysis of
metabolic and signal transduction networks. Because the number of experimentally
determined structures for protein-protein and protein ligand
complexes is still quite small, methods for computational prediction of
protein-protein interaction sites are becoming increasingly important.
In this project,
classification and prediction of protein-protein interaction will be
studied. In the training phase of
analysis, and given a protein and the fact that it can form a complex with
another protein, the reliability of predictions of, say, which amino acid
residues are located in the interaction site and other structural
considerations will be investigated. In
these studies, different aspects of interaction sites, such as hydrophobicity and other residue features, sizes shapes,
solvent accessibility, and residue pairing preferences, will be examined. In a more sophisticated analysis, second-order
structure can also be taken into account.
Much of this project
will involve database investigation but will also utilize statistical testing
and classification methods.