Top: Estimate of method bias, lower is better. Networks were evaluated on standards derived from single BioGRID evidence types and compared to all edges from BioGRID. Increased or decreased evaluation scores (F1) suggests bias of the method toward a specific type of evidence. This plot shows the relative variance of F1 scores from subsets compared to the full dataset. Bottom: Cumulative method performance across all evaluations, F1 scores. For all three species as well as KEGG, BioGRID and YEASTRACT standard, the highest performing pruned network was selected for each method. The bar plot shows the sum of maximal F1 scores, shaded by species.