Figure 2

Examples and explanations of measures. This figure shows two node alignment clusters, $\mathfrak{B}\left(u\right)=\left\{u,o,q,s\right\}$ and $\mathfrak{B}\left(v\right)=\left\{v,p,r,t\right\}$, and a corresponding cluster of induced edges $\mathfrak{E}\left(u,v\right)=\left\{\left(o,p\right),\left(q,r\right)\right\}$. Note that not all aligned nodes induce edges in general; sand tin this example are assumed to not interact in any of the aligned PPINs. The pairs u:o, v:p, v:rand p:rall contribute to the number of aligned nodes with high functional similarity NFbut only u:o, v:pand v:rcount towards the scaffold centric measure NF _s . p:rdoes not contribute to NF _s as the pair does not contain V, the root node of the alignment cluster. Equally, q:sand v:tcontribute to the number of aligned homologous nodes NHwhile only v:tadds to NH _s , the scaffold centric version of the same measure. Each of the node alignment clusters contributes a count of 3 to NA _s , the number of nodes that are aligned to the scaffold nodes, and 6 to the number of all possible pairs of aligned nodes NA. Considering the conserved interaction (o, p), since both endpoints are functionally similar to the endpoints of the interaction between the scaffolding nodes spanning the respective alignment clusters (e.g. u, oand v, pare functionally similar), it contributes to EF _s (and also to EF). Note that u, v(and thus o, p) do not need to be functionally similar to each other. The fact that sand tare homologous does not lead to any contributions. The cluster of induced edges adds 2 to the total number of interactions aligned to the scaffold interactions EA _s and 3 to the same measure accounting for all possible pairs of aligned interactions EA. Even though there are three conserved interactions, this cluster contributes to EA-2only since the three edges belong to only two distinct species.