Figure 7
Workflow of SI-plot analysis. (a) The only input to the algorithm is the structure of a signaling network represented by an adjacency matrix with directions and signs. (b) The influence of node j on node i (Sij) is estimated using only the input adjacency matrix. The first step is the normalization of link weights, in which each link weight is determined by averaging the degrees of source and target nodes. The second step is the computation of influence based on the signal flow estimation algorithm. Sij consists of the local response of node j to the perturbation of node j (∂xj/∂bj) and the global response of node i to the change in node j (dxi/dxj). (c) The output is an SI-plot, which is obtained by plotting the influence against SPLO. Promising control targets can be discovered by prioritizing the nodes in the SI-plot according to certain criteria such as a set of the top-k or top-n% most influential nodes on the output in each SPLO group.
