The Establishment Of Ruminant Cecum Metagenomics And The Development Of Machine Learning Microbial Prediction Model

Ruminants play an important role in agricultural safety,and their meat and milk are a major source of protein for humans.Therefore,to clarify the mechanism behind how ruminants convert low-value lignocellulosic plant material into high-value animal protein,we may be able to produce more food while using fewer resources,a key aim of improving global food security.As an important fermentation organ,cecum can be degraded by microorganisms using structural carbohydrates.Previous studies were mainly based on 16 S r DNA analysis,the species classification was vague and could not provide accurate microbial function information.Based on this,our research is based on metagenomics to establish the first goat ruminant cecum reference microbial genome catalog,and elucidated the multiple evolutionary strategies of bacteria to decompose plant polysaccharides from a function genomic perspective.In addition,collected and assemble the non-ruminant cecum microbial genomes,and builded a ruminant cecum microbial prediction model based on machine learning methods.The main results are as follows:1.A total of 567 non-redundant genomes were obtained by using 150 G data of 15 sheep caecum samples,of which 30.3% were assembled close to complete high-quality genes.In this study,the cecal microbial genome was annotated and a genome-wide phylogenetic tree was established.Bacteroides and Firmicutes are the dominant flora of cecal microorganisms.The 554 microbial genomes are less than 95% similar to the existing public data tree genomes,indicating that there are a wide range of potential new microbial species in the cecum of goats,and it provides a reference genome and classification basis for subsequent cecal microbial genome research.2.Cecal microorganisms have multiple strategies to degrade complex carbohydrates.First,the cecal microorganisms possess a large number of abundant and novel enzymes.Carbohydrate enzyme analysis showed that among the 893,086 proteins of goat cecal microorganisms,49,738 were annotated with at least one carbohydrate enzyme activity module gene;secondly,the cecal microbes have a polysaccharide utilization site carbohydrate coordinated utilization system,and the polysaccharides of 567 goat cecal intestinal microbial genomes are analyzed.After window scanning and statistics using the sites,it was found that all polysaccharide utilization sites appeared in the genome of Bacteroides phylum.Among the 167 Bacteroides genomes,113 genomes have a polysaccharide utilization structure,and 86% of them have more than one polysaccharide utilization site.By constructing a goat cecal microbial carbohydrate metabolism network,it is found that the entire microbial flora has diverse substrate types,reflecting the diversity of the host’s diet and leading to changes in microbial carbohydrate degradation strategies.There are functional redundancy in the cecal microbes,and a variety of microorganisms are It can degrade the same substrate;analysis found that the cecal microorganisms possess 23 quorum sensing signal synthetase enzymes,which can generate 13 different quorum sensing signals,and the various ways of microbial communication facilitate inter-species cooperative degradationIn this study,567 ruminant cecal genomes and 1,710 non-ruminant genomes from pigs and chickens were collected and integrated.After functional annotation,the machine learning model was developed using the Random Forest Bagging ensemble algorithm,based on the highly complex data in the orthologous egg identification function matrix.By creating random subsets and using these subsets to build decision trees to prevent over-fitting from happening,and using supervised algorithms to predict the source of a single gut microbial genome based on its host source.The accuracy of the model is 90.80% and the receiver operating characteristic(ROC)curve value is 0.97,which proves that the machine learning model has good specificity and sensitivity and can accurately distinguish the cecal intestinal genome of ruminants and non-ruminants.In summary,we pioneered the establishment of a ruminant cecal microbial reference genome,and conducted a detailed analysis of microbial classification and functional research.In practical applications,our results make it possible to precisely regulate the cecal flora to promote the growth and development of the host,and the various new enzymes identified provide technical support for subsequent industrial use and development of biological resources.In addition,it is the first time to apply machine learning to the analysis of animal flora to establish a host prediction model based on the function of the flora,providing a new perspective and idea for clarifying the functional advantages of ruminant flora to guide animal feeding..