Biology Reference
In-Depth Information
in a cell, and they capture the majority of direct connections that can
occur between proteins.
Of course, current protein-protein interaction networks still repre-
sent only a much-simplified view of the true connectivity of proteins
inside a cell. For the most part, the networks are not yet capable of
describing any spatial or temporal restrictions for protein interactions
(i.e. they ignore the fact that not all possible protein-protein interac-
tions actually take place in all cells of an organism and at all time
points). The networks are also not annotated with much detail regard-
ing binding energies, interaction stochiometry, or the spatial arrange-
ment of the proteins. They are also vastly incomplete — current
knowledge of protein-protein interactions is limited to a few model
organisms, and experimental measurements are often done under a lim-
ited set of laboratory conditions where many specific interactions may
not be forming.
Nevertheless, even the crude protein-protein interaction networks of
today represent a very powerful way of integrating and communicating
functional information about a cell's proteome. These networks provide
an exquisitely intuitive, concise visualization of complicated relationships,
and they can be used as a platform to navigate through entire proteomes
interactively. Many additional information items can be mapped onto a
protein network, and the networks can be used to place most types of
experimental data into context. Apart from visualization and data min-
ing, protein interaction networks can also help to address fundamental
questions about cell biology and evolution. For example, hierarchical
clustering algorithms, when applied to protein networks, can potentially
reveal “functional modules” or (sub)complexes forming molecular
machines, and can thus help to describe the overall functional organiza-
tion of a given proteome.
2
Likewise, graph-theoretical analysis of the
topology of interaction networks is used as a tool to search for funda-
mental design principles of evolution.
3
2. Experimental Protein Interaction Data
Protein-protein interactions cannot yet be reliably predicted
de novo
by computational means, although many promising avenues are using
