Essential Protein Prediction Method Based On Dynamic Network Segmentation And Multi-biological Information

As essential substances in protein interactions,essential proteins often interact with multiple proteins.Once the essential proteins in organisms is missing,it may cause the loss of some functions of organisms and then lead to the death of organisms.Therefore,how to accurately and efficiently identify essential proteins is of great significance to the study of life sciences.Although the traditional method of biological experiments to verify key proteins has high accuracy,it also has some defects such as long experiment period and high manpower and material cost.As computational science develops,a large number of essential protein identification methods based on computation have emerged at home and abroad.It includes classical methods based on PPI networks topological features,however,traditional method of computing centrality only focuses on the topological structure of PPI network,and PPI network itself also has false positive and false negative problems,which will greatly affect the recognition rate of essential proteins.Then some scholars propose a dynamic network construction method based on dynamic threshold on the basis of static network,which made up for the limitation of static network to some extent and improved the recognition performance of essential proteins.However,its recognition accuracy still has bottlenecks.In view of the above problems,the main work contents of this paper are as follows:(1)This paper introduces the concept of periodic gene expression and proposes a dynamic network segmentation method.In this method,the gene "active" expression matrix is constructed,and the segmented "active" expression matrix is used to act on the protein-protein interaction network to form the protein periodic subnetwork.Finally,the importance of the protein nodes in the network is measured by integrating each protein periodic subnetwork.The experimental results shows that the method could effectively improve the prediction rate of essential proteins in yeast,E.coli and human bladder data.(2)Based on the conserved and functional enrichment of essential proteins,this paper proposes a protein domain enrichment fraction,which is used to measure the enrichment degree of protein domains in essential proteins,and a MBM model is proposed by using Multi-layer perceptron(MLP).This model combines many kinds of biological information characteristics,such as domain enrichment score,subcellular localization,protein complex,the topology and composition features of static PPI network,as well as networks topological and synthetic features of dynamic network after segmentation proposed.The experimental results show that MBM model can identify more essential proteins.