Figure 7: Performance of prediction sets based on the complex sets M, C, N, and N+ derived from the different λ. The tradeoff threshold 0.7 almost gets good performance based on the number, statistical likelihood ratio , and percentage hit in BioGRID of the PPI predictions.