Fig. 3
From: A two-step framework for inferring direct protein-protein interaction network from AP-MS data
The performance comparison on the Krogan data set when SA, PE, Hart and DC are used as the scoring method in the initial phase and BGS, PCA and Y2H are used as the reference set. To obtain a filtered network, we use both ND and Silencer as the indirect interaction removal method in the second phase. For each reference set and scoring method, we report a set of top-ranked interactions (x-axis) to check how many interactions are contained in the the reference set (y-axis)