Research On Multi-omics Data Integration Algorithm Of Rice Infected By Magnaporthe Oryzae Based On Machine Learning

Rice is one of the most important food crops in the world.Its yield and quality greatly affect people’s life.However,rice is easily infected by Magnaporthe oryzae,resulting in rice blast,which will greatly reduce the yield of rice.Thus it can be seen that finding a long-term and broad-spectrum control method for rice blast is very important.Research found that small RNAs（sRNAs）of plant fungal pathogens can transboundary regulate host plants to facilitate their infection.Therefore,integrating M.oryzae-rice multi-omics data and exploring M.oryzae key pathogenicity-associated sRNAs,which play an important cross-kingdom regulatory role in rice infection,are very important to controll rice blast.However,currently there are few studies on how to integrate M.oryzae-rice data and mine key pathogenicity-associated sRNAs of rice blast infection.In this paper,the M.oryzae-rice multi-omics network（MRMO）was established by integrating the multi-omics data of M.oryzae infecting rice.On this network,the NRMRMO and NL-MRMO methods proposed in this paper were used to mine the key pathogenicity-associated factors of M.oryzae.Firstly,M.oryzae transcriptomics data,proteomics data,rice genomics data and proteomics data were obtained from biological databases.According to the target gene prediction and protein interaction network establishment,the MRMO network was constructed.Then the NR-MRMO and NLMRMO methods are applied to the network respectively.The specific contents of the two methods are as follows:1.NR-MRMO is a data analysis model for the MRMO network,which combines the node2 vec graph embedding algorithm and the random walk with restart algorithm.Firstly,the node2 vec is used to obtain the low dimensional vector representation of nodes in the MRMO network,and then the cosine value of edges in the MRMO network is calculated as the weight according to the embedding results.Finally,the weighted MRMO network is put into the random walk with restart algorithm to obtain the relationship between M.oryzae sRNAs and rice genes.2.NL-MRMO is also a data analysis model combining the node2 vec algorithm and the Light GBM algorithm for the MRMO network.First according to the low-dimensional vector representation of the nodes in the MRMO network obtained by node2 vec,the relation vector representation of the edges in the MRMO network is calculated.Then taking the relation vectors as the input of Light GBM algorithm,the relationship between M.oryzae sRNAs and rice genes was obtained.The relationship between M.oryzae sRNAs and rice genes was obtained in both models,and 34 and 21 key pathogenicity-associated sRNAs of M.oryzae were screened respectively,including 10 overlapping sRNAs.GO/KEGG analyzed the genes corresponding to the sRNAs.The results showed that M.oryzae sRNAs can transboundary regulate the DNA catabolic process of rice genes and other biological processes,related to the pathogenic process of M.oryzae.Map the relationship between key pathogenicity-associated sRNAs and rice genes to show the many-to-many regulatory relationship between sRNAs and genes.Firstly,this paper proposed an integration method of the MRMO network.Then,based on this integrated network,45 pathogenicity-associated sRNA regulated by M.oryzae were mined by using novel NR-MRMO and NL-MRMO methods.Knocking out these key pathogenicity-associated sRNAs of M.oryzae can effectively reduce or eliminate the harm of M.oryzae.The multi-omics integration method in this paper provides a new idea for the integration analysis of plant-fungus multi-omics.This study reveals the regulatory role of sRNAs in the process of rice blast infection,which will help us better understand the interaction mechanism between rice blast and its host rice.In addition,this study looks for the pathogenicity-associated factors of rice blast from the perspective of computer,which is more friendly to the environment and lower cost than the commonly used chemical control methods,and provides a new reference scheme for the control of plant fungal diseases.