Figure 1

The WGD transcription network of yeast. Top-left panel: cartoon of the duplication-divergence model. Following a duplication, TF binding sites of duplicate promoters are initially identical. Evolution can retain memory of this initial condition, producing links in the network with consistent statistical patterns compared to homology classes [10, 19]. Mid-left panel: scheme of all the possible DSs originating from an ancestral interaction. DSs are listed by number of links and node degree; DSs that are identical except for exchanges of the duplicate copies are grouped together, as 'old' and 'new' nodes cannot be distinguished in WGD. Each group is associated to a color and a greek letter. Bottom-left panel: scheme of the analysis procedure. First, the nodes of each duplication age class in the network and corresponding links are collapsed, in order to estimate the ancestral network. Subsequently, the DSs associated to each ancestral link to annotate the links. This operation is a coloring of the links, where each color corresponds to a DS. Fully disconnected η-type DSs are ignored. Right panel: the yeast "WGD network" of all nontrivial DSs (having two or more links). Although ζ type DSs, which make up 94.51% of the network, are not represented, the network is strikingly connected. Red (β, 1.1%) and blue (γ, 4.2%) DSs are heavily dominant. Furthermore, links of the same color are clustered and arranged in a hierarchical fashion (where red precedes blue). The occurrence of the other DSs is: α 0.041%, δ 0.049%, ε 0.066%.