Learned likelihood parameters, expressed as likelihood ratios, for (a) yeast, (b) human. For PPI data, interacting proteins are likelier to be co-complex when they have more shared neighbors. For L2-PPI data, where protein pairs are not reported to interact yet share many interaction partners, proteins are also likelier to be co-complex when they have more shared neighbors, but at lower likelihood ratios than corresponding PPI values. For STRING data, protein pairs with predicted functional associations are very likely to be co-complex when the prediction score is high; at low scores, protein pairs are not much likelier to be co-complex. For PubMed data, protein pairs that co-occur in literature, even infrequently, are already much likelier to be co-complex; however, pairs that co-occur more frequently in literature are not any more likelier to be co-complex compared to pairs that co-occur less frequently.
Yong et al. BMC Systems Biology 2012 6(Suppl 2):S13 doi:10.1186/1752-0509-6-S2-S13