Association Analysis Of Root Microbiome Data And Plant Yield Based On Deep Learning Model

The Microbiome refers to all genes and genomes from microorganisms including bacteria,fungus,archaea,and some lower eukaryotes.Plant microorganisms colonize in plant organs and plant surfaces.Some specific microbiome only colonize in specific species or their specific organs.For example,rhizosphere microorganism communities are only formed under the influence of plant secretions in narrow areas of plant root surfaces.Microbiome play an important role in plant growth and in protecting plants from stress,while environmental and plant genetic and metabolic regulation can shape and influence microbiome.The interaction between microbiome and their host,such as genetic material and metabolic exchange,has become a hot research area.As high-throughput sequencing technologies and the development of the bioinformatics analysis methods,using omics data to study microbiome and the relationship between microbiome with their host has become the trend of the present study,it is no longer restricted to the study of certain known microbes,and more easily to find important genetic information.By studying the composition of the core microbiome and their interactions with plant hosts,we can utilized the knowledge of microbiology to predict host phenotypes,thereby help plants achieve increased yield and stress resistance under biotic and abiotic stresses.The deep learning methods have been increasingly applied to learn the microbiome data due to their powerful strength of handling the complex,sparse,noisy,and high-dimensional data.Unraveling the association between microbiome and host phenotype can illustrate the effect of microbiome on host and then guide the agriculture management.In this study,we summarized the analytic strategies and critical steps in the microbiome data processing.We discussed the accessible usage of the models in plant-microbiome correlation analysis and prediction task,and clarify how the models improve the performances of the analysis pipeline.We also introduce and summarize the superiority of deep learning methods in modeling and interpretation for association research.We adopt the convolution neural network in predicting the productivity of milled based on abundance feature map.The precision of model reaches 70%.This attempt provided a new angle for plants microbiome data to conduct the association study.This study surveyed the analysis strategy of plant related microbial groups of data,and summarized the correlation and prediction analysis tasks of some of the key steps in deep learning model,put forward the problems that exist in the modeling,our work provided references for deep learning methods in the study of host phenotype correlation analysis application.At the same time,the convolutional neural network was adopted in this study to construct and predict the yield capacity of millet based on the characteristic data of plant microbial abundance,which also provided a new paradigm for the study of host phenotype association with plant microbiome data.