Figure 5

Varying the Co-occurrence Threshold. Relative performance of Functional Flow when varying the threshold used to define the co-occurrence interaction set. Shown is the number of true positives (TP) when the scoring threshold is set to yield 100 false positives (FP) (y axis). The values of the x-axis denote instances of Functional Flow on graphs combining PPI and the interaction sets for each corresponding setting of the co-occurrence threshold (x = -1 shows PPI ONLY and x = 0–9 denote PPI plus the datasets obtained using thresholds 0.0 to 0.9). The lines are annotated to denote the MUT, HYG and ACF metrics. The best and worst performers respectively, over all co-occurrence measure and all thresholds, are shown in parentheses below the plot title. These combinations appear as Best and Worst in Figures 2 and 3.