Figure 4. The framework of network fingerprint extraction algorithm.

(A) Network merging. The networks to be compared, G₁ and G₂, were merged into a virtual integrated fully connected graph G^*. (B) Functional clustering. The edges of G^* were weighted by assigning the functional similarity between the interacting nodes based on GO. The AP clustering algorithm was employed to identify the functional modules (represented as different colors) of G^*. Nodes in a functional module came from G₁ and G₂ (e.g., in the red cluster, nodes f, g, and h came from G₁, and nodes E and G came from G₂). (C) Similarity scoring. The functional similarity between two networks was calculated based on the clustering results.