The Classification Of Node In Biological Network Based On Network Representation Learning

With the completion of human genome sequencing,the "post-gene" era has opened,and the focus of the scientific community on the function of life has gradually shifted from single-objective analysis to omic analysis.In the analysis of omics,we mainly use a large amount of accumulated data of biology,and computer technology to analyze from a sys-tematic point of view,which greatly improves the efficiency of the research.At the same time,computer technology can model biological processes,guide biological experiments based on models and large amounts of data,furthermore improve the efficiency of biolog-ical experiments.Bioinformatics,as an interdisciplinary subject,combines life science with computer science,and puts forward new ideas for future biological research.Bioin-formatics emphasizes the discovery and interpretation of related life mechanisms from a systematic point of view,which is also the inevitable trend of future research.In this pa-per,we studied the classification of nodes in biological networks,classification of nodes is helpful for us to mine the role of nodes in the network to infer their role in life activities.In addition,node classification can also infer the functions of unknown nodes,which has guiding significance for the next biological experiment.Node classification can help us understand the importance of nodes in the network and the distribution of nodes in the network.In this paper,we mainly accomplish the following two works:1)Node classi-fication in gene networks.Genes cooperate with each other to fulfill their life functions through interaction,which forms a gene network.Classification of genes in gene networks can help us understand the role of genes in life activities and the importance of genes.The idea of information flow is introduced into gene networks through structural hole,and we use network representation learning in HumanNet based on it.Nodes are classified based on the results of representation learning.We analyzed different types of nodes and veri-fied that different types of genes play different roles in life activities.2)We design a new GAN-based network representation learning algorithm,walkGAN.The algorithm is ap-plied to the disease-gene network to classify disease based on the results of representing learning.We compare our method with some classical network representation learning algorithms.The experimental results show that the performance of our algorithm is much better than other algorithms.