Research On Reconstruction Methods Of Gene Expression Data Regulation Network

Inferring gene regulatory networks plays a vital part in Bioinformatics,as it provides a new analytical tool for the mechanism study and function analysis of biological organisms.Currently there are a host of methods for inferring regulatory networks.However,due to the fact that there are a large number of potential direct or indirect regulatory relationships for gene data but a small number of samples of gene expression data,the recognition rates for most existing methods are relatively low.Inferring gene regulatory networks from gene expression data remains a challenge in Systems Biology.This research focuses on the reconstruction of the gene regulation network methods and the corresponding system.In this thesis,different gene regulatory networks were developed respectively for different types of gene expression data to improve the accuracy and reliability of the algorithm,and on the basis,a gene regulation network inference system was built.The main work and contributions of this thesis are as followings:1.Propose an inference method for static gene data regulation network: a gene regulatory network inference method for the steady gene expression data,which includes the entire process of decomposition,learning and fusion,is proposed.Randomization and regularization are introduced to solve the problem of model overfitting,and normalization and statistical approaches are introduced to solve the problem of different primary weight in modeling.As a result,the accuracy and stability of the inference model are improved.2.Propose a priori-fused boosting method for gene regulatory network inference:we present a novel method,namely priori-fused boosting network inference method(PFBNet),to infer GRNs from time-series expression data by using the non-linear model of Boosting and the prior information(e.g.,the knockout data)fusion scheme.Considering the accumulated information of gene expression at previous time points,other types of data were integrated into the model as prior information,which greatly improves the performance of the prediction model.3.Develop a gene regulation network inference prototype system: a regulatory network inference system,which integrates the static and time-series gene expression data,is developed in this research.This system for the first time adopts the integrated framework algorithm based on prior fusion to infer online gene regulatory networks,and visualizes the regulatory networks obtained by the algorithm,achieving a better effect of online network construction and analysis.It is a common goal in Bioinformatics to accurately and efficiently reconstruct networks using different types of gene expression data.In this thesis,to address the shortcomings of the conventional methods of gene regulation network construction,optimization methods for static and time-series gene expression data are proposed,which significantly improves the accuracy and reliability of gene regulation network reconstruction algorithm.