High-order Interaction Analysis Of Microbial Network Based On Weighted Motifs And Logical Networks

Microorganisms play an important part in the biological process of the human body;they rarely live in isolation,but instead form a complex ecosystem interaction.Understanding the relationship between community composition and function is a major challenge in microbial ecology.It is of great significance for the management of natural microbial communities and the design of synthetic communities.At present,there are few studies on microbial motifs and higher-order modules,so the research on higher-order modules of the microbial network has become a hot topic.Biological networks are modular,while the traditional methods based on network motifs often neglect the effects of the strong or weak interactions between microbe nodes on the function of the network.The higher order relationship of microorganism is regarded as a hyperedge,but the weight between hyperedges is not considered in the algorithm based on hypergraph clustering.Therefore,in this paper,the main research efforts and innovations are as follows:First,high-order microbial organization based on weighted motifs.Based on the microbial time series data after taking antibiotics,a dynamic microbial network is constructed by using the Dynamic Bayesian Network algorithm,and the normalized Bayesian Factor represents the weight of microbial network.In this paper,considering the strong or weak relationship of the node interaction in motifs,the definitions of weighted networks and their corresponding weighted motifs are given,and a method based on the weighted directed motif is proposed to analyze higher order organization of the weighted microbial network.The experimental results show that with the information of high-order relation strong and weak interaction,the partially weighted motif can obtain better mathematical and theoretical clusters than the unweighted one.and the microbial module can be investigated more accurately.Second,the transitive hypergraph clustering algorithm based on logical networks.The traditional microbial network construction and its network structure are based on paired interactions.However,the microorganisms vary a lot,and the diversity of communities depends a large extent on the stability of higher-order interactions.In this paper,based on the microbial abundance data at different sites of the human body,the ternary groups of microorganisms satisfying the logical relationship are calculated for each part of the human body.Then the microbial hypergraph is obtained by constructing the hyperedges of the microbial triplets.Besides,considering the structure of the triple to incorporate the information of the hyperedge similarity.So that to propose a transitive hypergraph clustering algorithm,and the problem of unknown clustering number is solved by maximizing the modularity.Finally,according to the index of joint entropy,total correlation and modularity,it is measured the performance of high-order module mining based on transitive hypergraph clustering algorithm,and proved that this method is effective and feasible.