Figure 3From: Predicting protein linkages in bacteria: Which method is best depends on taskFunction Prediction Cross-evaluation of protein linkages for each method (PP, RS and GN): UNIF: predicted function is sampled uniformly at random from the set of categories (KEGG 4 categories, COG 19 categories, KEGG 19 subcategories); MAJOR: predicted function is the majority assignment to immediate neighbors (ties are broken randomly).Back to article page