|
D.Phil. Research Proposal:
Our current focus is on protein-protein interaction networks, which represent experimentally observed physical binding interactions between proteins in a cell (these are collectively referred to as the interactome). The development of high-throughput screening techniques has led to the compilation of large interaction datasets, in particular for yeast. The two major experimental methods used are yeast two-hybrid (Y2H) screening [12,13,24,36], and tandem affinity purification followed by mass spectrometry (TAP/MS) [15,23]. The quality and reliability of available datasets is a major issue; recent studies [37] indicate that the properties of Y2H and TAP/MS data differ substantially, with the latter mostly picking up interactions that are part of protein complexes, whilst Y2H is better at capturing more transient binary interactions. Consequently, interaction networks constructed from the two kinds of data also tend to have different characteristics, and one of the key problems in this area is how to combine insights from the different experimental sources in order to obtain a comprehensive picture of the interactome’s organisation.
One of the major theoretical approaches to modelling sets of interacting elements in various domains has been to look at them as graphs or networks [30]. Here the elements are represented by nodes, and interactions are represented by links between nodes. In particular, for protein interaction networks, the usual method is to have a node for each protein, and place a link between all pairs of interacting proteins. Typically, no direction or weight is attached to these links; thus the interactome is generally modelled as an unweighted, undirected network. Various properties can be studied for such networks - a particularly relevant one in this context is that of community structure [11, 17]. Roughly speaking, a community (also sometimes referred to as a cluster or module) is a set of nodes with a higher than expected number of links amongst them, as compared to links to nodes outside the community. Many real-world net-works have been shown to possess significant community structure [30], as compared to random networks with the same distribution of node degrees. In particular, several studies have pointed towards modular organisation of the proteome [16,20,26], with densely interacting groups of proteins being responsible for specific functions and processes. Community structure has also been seen in metabolic networks; a study by Guimer`a and Amaral [19] showed that metabolites could be assigned distinct roles in the network by means of a topological analysis based on communities. 我要提问
|