A Linear Regression Model Of Gene Regulatory Networks And The Robustness Analysis

Gene Regulatory Networks (GRN for short) are the networks which show thecomplex regulatory relationships between multiple genes, the relations of mutualregulations between genes affect various physiological activities and actions. The studyof the GRN starting from the interaction between genes, revealing the essence of thephenomenon of complex life organisms, is an important part of studying of genefunction in genomics. The study of the GRN has important theoretical and practicalsignificance in applications, has become a hot topic of research in bioinformatics.This paper studies the construction of the GRN based on the linear model. Firstly,feasibility of evaluating the GRN with the linear model is studied, we use the RAF dataset to analysis the gene expression data to demonstrate whether the linear relationshipsexist, the method PCA is used in order to have an in-depth research. Secondly, the linearGRN with the LARS algorithm for getting the Lasso path is built. Finally, therobustness to construct the GRN with a linear model is comprehensively analyzed inthree aspects which contain noise adding, data losing and samples reducing respectively.