Protein coverage summary and number of predicted complexes.

PPI datasets used in this study contain 1622 and 3006 proteins for yeast and human respectively [23]. In yeast, all of these proteins can be found in the abundance datasets [3, 16], while in human only 88% of proteins in the PPI network have data on protein abundances [10]. More information is known about domains in yeast than in human [21]. The final yeast model contains 84% of the interactions and 91% of proteins from the initial PPI network. The human model contains 67% of the interactions and 76% of the proteins of the PPI network (Table 1). The missing proteins and interactions are due to the lack of DDI interactions or shared GO function between two proteins of the same interaction (S1 Text for additional details about model generation). The whole pipeline generated 657 complexes in yeast from which 248 do not match any known complexes [9]. From the human data we predicted 1158 complexes and 890 of these cannot be associated to any CORUM complex.
Protein coverage summary and number of predicted complexes.