Protein-Protein Interaction Temporal Network Model And Its Dynamics

The protein-protein interaction network(Protein-protein interactions,PPI)is the basis for cells to metabolic activities.The PPI network is a biological network that describes the activities of proteins in cells.,which its research has a positive guiding role in understanding the biological processes,revealing the mechanism of disease occurrence,and developing new drugs.At present,the research of PPI network has changed from the single source data static model to the multi-source data dynamic model and the integration of static PPI data and time series gene expression data to construct a dynamic PPI network is an important research direction.In this thesis,the following aspects are carried out.The difficulty in constructing a dynamic PPI network lies in the binary processing of gene expression data.The existing methods can not fully retain the interaction information of the static PPI network,and even filter out the interactions which are related to the essential proteins,which affects the identification of essential proteins to some extent.In order to solve this problem,a dynamic PPI network construction method which is based on the minimum full coverage is proposed.At first,the expression levels of each protein at each time point are sorted,and based on a low-to-high order,using the subtraction method to remove all time points that do not affect the full coverage of the interaction,the shortest activation time sequence that can cover the interaction of all static PPI networks is obtained,and a dynamic PPI time series network is obtained by the static PPI network projection according to the activation time sequence.On this basis,the characteristics of dynamic PPI network such as node degree distribution,node activity,edge activity,and hub node distribution are discussed,which provides a basis for understanding the dynamic characteristics of PPI network from the perspective of timing.A weighted clustering coefficient' centrality measurement method is proposed by calculating all interacting edge clustering coefficients in the PPI network at each time point at first,Then obtaining all interactive edge clustering coefficients of the dynamic PPI by weighting,at last summing all nodes' clustering coefficient.The weights at each time point are optimized through intelligent algorithms.the experiments show that the method by a weighted clustering coefficient'centrality measurement has a higher recognition rate of the essential proteins than the existing centrality measurement method;the PPI network constructed by the the minimum all-overlapping method which has a higher essential protein recognition rate than the network constructed by other methods.The minimum all-overlapping method and the weighted clustering coefficient' centrality measurement method which proposed in this paper can improve the correct recognition rate of the essential proteins effectively and provide a new exploration approach for PPI network research.