Figure 2.

One-mode projection of the gene–function network. A gene–function network can be turned into a functional gene network through a one-mode projection, which connects two genes if they are next-nearest neighbours in the gene–function network. This process is not one-to-one as is evident from the two examples given, and the functional gene network therefore contains less information as it disregards which pathways two adjacent genes share. The gene–function network can be viewed as a decomposition of a functional gene network because the latter can be constructed by multiplying the adjacency matrix of the gene–function network with its transpose. This bipartite network separates the gene and function connectivity into two separate distributions, and therefore gives a more detailed description of gene functional association.