Across species: How does inferring interactions from the source-species interactome compare to inferring interactions from randomised versions of the source-species interactome?

We give the ratio of the fraction of correct inferences from the real interaction data compared to randomly rewired data for the reciprocal-hits homologs. (The number of interactions in which each protein participates is preserved in the randomization.) The numbers in parentheses give the standard deviations over rewirings.