Construction of Uncertain Protein-Protein Interaction Networks and Its Applications

Bihai Zhao, Jianxin Wang, Fang-Xiang Wu, Yi Pan

Abstract: Recent developments in experiments have resulted in the publication of many high-quality, large-scale protein-protein interaction (PPI) data. Unfortunately, a significant proportion of PPI networks have been found to contain false positives, which have negative effects on the further research of PPI networks. We construct an uncertain protein-protein interaction (UPPI) network, in which each protein-protein interaction is assigned with an existence probability using the topology of the PPI network solely. Based on the uncertainty theory, we propose the concept of expected density to assess the density degree of a subgraph, the concept of the relative degree to describe the relationship between a protein and a subgraph in a UPPI network. To verify the effectiveness of the UPPI network, we propose a novel complex prediction method named CPUT (Complex Prediction based on Uncertainty Theory). In CPUT, the expected density combined with the absolute degree is used to determine whether a mined subgraph from the UPPI network can be represented as a core component with high cohesion and low coupling while the relative degree is the criterion of binding an attachment protein to a core component to form a complex. We employ CPUT and the existing competitive algorithms on two yeast PPI networks. Experimental results indicate that CPUT performs significantly better than the state-of-the-art methods.