The Construction Of Dynamic Protein Interaction Networks And Protein Complex Detection

Proteomics is the most exciting frontier in life science. It becomes one of the hottest researches to systematically analyze and comprehensively understand protein as the undertaker of the life activities through the study of protein structure, function, and interaction. In particular, identifying protein complex from the protein interaction network has important significance on revealing the structure of protein network, predicting protein function, explaining particular biological process.The static protein interaction network, in which the interactions are accumulated in different conditions and time, cannot reflect the real dynamic protein interaction network in cell, and therefore has an influence on the accuracy of protein complex prediction. Focused on this shortage of static protein interaction network, this paper propose a method to construct dynamic protein interaction network based on static protein interaction network and3-sigma principle which is used to identify the active time points for each protein. To evaluate the performance of this dynamic protein interaction network, MCL algorithm is applied on it to identify protein complex. Compared with the performance of other two protein interaction network, the results show our dynamic protein interaction network is a network which can more precisely reflect the dynamics in real protein interaction network, and it makes predicting accurately protein complexes with certain biological meanings possible.The protein complex’s formation and inherent biological properties are neglected in algorithms which focus on topology and function similarity to predict protein complex. Based on the just-in-time mechanism followed by protein complex formation and protein activity, this paper presents a method to refine the protein complexes predicted by the existing methods. The comparison between the protein complexes before refinement and that after refinement shows that significant improvements on evaluation metrics are gained after refinement. The results show the refining method can capture the right inherent properties of protein complex, and the biological properties play a vital role in protein complex detection.The method of dynamic protein interaction network construction and protein complex refinement method proposed by this paper from different angles efficiently solve some problems in static protein interaction networks and protein complex identification, make the predicted protein complex more meaningful, and can provide some references for biologists in their biochemical experiments.