Figure 5

Enrichment over random site predictions. Using highest peak only, for yeast and human (a), best of one, two, three peaks peaks in yeast (b) and human (c). For both species, we predict significantly more sites close to the experimentally-confirmed site than random. Yeast performs slightly better due to the much greater coverage of the species interactome by the PIPE known interactions database yeast (36,000 known pairs out of estimated ~50,000 [30]) compared to human (40,000 known pairs but much larger interactome).