Research On Key Protein Identification Method Based On Fusion Dynamic PPI Network

Essential proteins play an indispensable role in supporting the life of an organism.Lack of Essential protein will result in cell death or stop reproducing.Identifying essential protein will help us understand the basic mechanisms of cell life.The essential genes and its coding proteins of bacteria are potential drug targets of some diseases.Therefore,identification of essential proteins has great significance both in the study of life sciences and the design and production of drug.In recent years,some methods of identifying essential proteins based on static protein-protein interaction(PPI)network have been proposed.However,these methods have ignored the fact that essential proteins play essential roles under certain conditions.According to this feature,the main works of this dissertation are as follows:(1)We systemize and introduces existing methods of dynamic network construction.The performance of dynamic network and static network in essential protein identification was compared and analyzed using classical centrality methods.The experimental results indicated that using dynamic networks can effectively improve the accuracy of essential protein identification methods based on network topology.(2)In this study,we have designed an essential proteins identification method based on fusing dynamic PPI networks.First,considering essential proteins interact together in the form of function modules in different dynamic PPI network,we weighted the subnets of dynamic network by the edge clustering coefficient algorithm.Then,for dynamic PPI network,we improved the method named Similarity Network Fusion to fusing subnets of dynamic PPI network.And a final network is constructed using this method.We use the random walk method based on the final network to capture the global information of each protein,and use the interaction frequency entropy to measure the local information of each protein in the final network.Finally,combined with the global information and local information,we designed this method called FDP.And FDP was compared with other methods of essential protein identification.From the result,FDP owned a better performance on essential protein identification.(3)In the further study,we find that FDP algorithm still has room for improvement.So we improved FDP algorithm and named it FDP-LAC.FDP-LAC,like FDP,uses random walk algorithm to capture the global information of each protein in the fused network.Then,we pruned the fused network,and the weighted LAC algorithm is used to calculate the local score of each protein on the pruned network.The FDP-LAC is designed based on the global and local scores of each protein.Finally,we compared the experimental results of FDP and FDP-LAC,and the results show that FDP-LAC is better than FDP about essential proteins identification.