Method For Constructing And Analyzing Biological Network Based On EQTL Data

eQTL analysis aims to identify genetic variations in cells or tissues that affect the expression of one or more genes,and is an important tool to reveal the relationship between genetic variation and gene phenotype.At present,most researches are based on eQTL bipartite network structure to mine modules to study the regulatory relationship of multiple SNPs to multiple genes and find modules that are significantly related to biological processes.However,due to the lack of gene interaction relationship,this bipartite network often leads to incomplete analysis and large errors in the results.Adding the gene interaction relationship to the eQTL bipartite network can significantly enrich the information in the network and make the results more reliable.In addition,the research of network embedding method simplifies the module mining problem in biological networks to use machine learning algorithms to identify network modules,so that the problem is no longer limited to the propagation algorithm based on seed set.In this paper,we propose a method for constructing eQTL networks based on gene interaction relationships and identify modules based on network representations.First,this paper calculates mediation effects based on linear regression models and non-parametric permutation test methods,and proposes methods for adaptively selecting the number of permutation tests and using GPD to approximate P-values.These methods can improve the speed and ability of detecting mediation effects.Then,this paper proposes to build a more comprehensive and rich eQTL network based on the eQTL bipartite network,combining the analysis results of mediation effects and the data of gene interaction relationship in public databases.Thirdly,this paper proposes an algorithm that uses the network embedding method to learn the node feature representation,uses hierarchical clustering to construct a tree,and uses dynamic pruning strategy to identify the modules in the tree.Finally,through the experiments on the ROSMAP data,the effectiveness of the mediation test algorithm(eQTLMAPT)is proved,and on the basis of this,the eQTL network is constructed by combining the gene regulation relationship data;through experiments in real-world networks and simulated networks,it proves the efficiency of the module mining algorithm(N2V-HC)based on network representation,and successfully mines meaningful modules on the eQTL network.