Fig. 4
From: Sensitivity analysis on protein-protein interaction networks through deep graph networks
An example of how sensitivity is predicted on a trained DGN. The user has to (1) select the subgraph containing the proteins of interest, here marked in yellow, (2) the input protein \(u_{in}\) to perturb, and the output protein \(u_{out}\) on which the sensitivity needs to be assessed. (3) The DGN model processes the induced graph and predicts whether \(u_{out}\) is sensitive to \(u_{in}\)