Figure 2

PPI data. From the human interactome, those PPIs (edges) are selected that have 1–1 relation with a publication; that is, the publication reports only one interaction, and that interaction is not reported by any other publication. The classification model is trained and evaluated using this 1–1 dataset. After evaluating the approach thus, all of the 1–1 dataset is used to train a new model which is then used to classify each of all of the edges in the interactome to identify high-impact edges. PPI network diagram was created with Cytoscape [29–31].