Figure 3

Weighted-and-Reverse-Swap (WaRSwap) treats the biological network as a composition of five possible source and target layers, The only layer that has self-loops is the TF→TF layer. Each layer is randomized by first permuting target in-degrees, then applying the core WaRSwap algorithm. The result is a randomized network that preserves target in-degree distribution with respect to each source-target type.