Essential Protein Prediction Based On Naive Bayes Algorithm And Bioinformation Fusion

Many functions of cell activities require the participation of proteins,such as the catalytic function of cells,the transport of metabolic products,and the reasonable regulation of metabolism in organisms.According to the difference of the importance of proteins in the life activities of organisms,proteins can be divided into essential proteins and non-essential proteins.The deletion of essential proteins will have a fatal effect on cells.The correct identification of essential proteins is of great practical significance for disease diagnosis.Therefore,it is of great significance to develop effective computational models to identify essential proteins.With the development of high-throughput biological experiment technology,it has become a research hotspot in the field of essential protein prediction by constructing the interaction network and carrying out the essential protein prediction research on the interaction network.However,there are relatively few known associations in the protein-protein association network and the protein-protein domain association network,resulting in very sparse association networks,which also negatively affects the prediction accuracy of essential proteins.At the same time,studies have shown that the biological information of a protein is potentially related to whether the protein is a key protein.Therefore,in this study,the known sparse protein interaction network was transformed into a relatively dense interaction network by using naive Bayes algorithm,so as to obtain more interaction relationships.Through the analysis of the above problems,the EPNBC key protein prediction model was proposed in this study.The main research contents are as follows:The model constructs the original protein interaction network(PPI)and the original protein-protein domain interaction network(PDI)respectively.By treating the PDI network with naive Bayes algorithm,a new protein connection network with more interactions is obtained.Based on the new interaction network and protein gene expression data,a weighted protein interaction network(WPPI)and a weighted proteinprotein domain interaction network(WPDI)were constructed,respectively.Through the interaction network of WPDI and WPPI,a heterogeneous weighted network with three kinds of connection relations can be further obtained.On this heterogeneous network,the final score of each protein was obtained through iteration using the improved Page Rank web page ranking algorithm.The Page Rank scores of proteins were ranked in descending order,and the prediction results of essential proteins were obtained.From the comparison results between EPNBC prediction model and dozens of other excellent essential protein prediction algorithms on different data sets,it can be concluded that EPNBC prediction algorithm has better prediction effect.