Figure 4

Selection of reference genome affects protein-protein interaction predictions.a) PPIs generated using Proteobacteria as the reference genome set. Red nodes are the nodes that are predicted when the reference genome is Proteobacteria and not predicted when the reference genome is All (confidence for Proteobacteria is r > 0.9 and for All is r > 0.8 since these two confidences give similiar PPV). b) PPIs generated using Motile as the reference genome. Green nodes are the nodes that are predicted when the reference genome is Motile and not predicted when the reference genome is All (confidence for Motile is r > 0.88 and for All is r > 0.8 since these two confidences give the same PPV) c) Two of the clusters that is predicted when the reference genome are Protebacteria not All. The proteins within this cluster are mainly membrane proteins that are Co-Conserved with hypothetical or unclassified proteins. d) Comparison of GGDEF clusters with two different type of reference selection. (All and Motile). When All is used as the reference genome set, the GGDEF domain is Co-Conserved with EAL domain and two component regulatory system proteins but when Motile is used as a reference genome set, GGDEF domain is Co-Conserved with EAL domain, two component regulatory system, regulator and proteins involved in metabolism. All these processes are known to contribute to the common phenomenon of biofilm formation.