Overall Performance. Overall performance of the scoring scheme on the MIPS, the Small Scale, and the Multiple Evidence gold standard data sets. The Random data set served as the gold standard negative. Each data point of the curves corresponds to a pair of true-positive and false-positive rates, defined as the fraction of GSP PPIs and GSN PPIs that achieved a score above a sliding threshold. The threshold ranged from 1 to 10-5 in this figure (values not shown). The area under the curve (AUC) is 84%, 86%, and 87% for the Small Scale, the Multiple Evidence, and the MIPS curve, respectively.