Prediction of the Directionality of Protein Interaction Network Based on Network Propagation Algorithm

Abstract

Biological signal transduction of protein-protein interactions (PPIs) network occupies an important position in life activities. To further understand the mechanism of life operation, an improved network propagation algorithm was proposed which, combined with protein semantic similarity and overlapping clustering algorithm, predicts the directionality of human PPIs network. As a result, 1 664 unoriented PPIs (form protein complexes) and 2 556 oriented PPIs (interactions involving signal transmission) were obtained, 95.79% of predicted unoriented PPIs have significant biological significance, the AUC of predicted oriented PPIs is 0.813. For human PPIs network, the algorithm realized better results for unoriented PPIs distinguishment and oriented PPIs judgement, which provides clues and ideas for further research on human PPIs networks.

Keywords

• network propagation
• human ppis network
• signal flow direction
• the weight of the ppi
• oriented ppis
• unoriented ppis