Correlation analysis as a dependency measures for inferring of time-lagged gene regulatory network

Abstract

One of the main aims of molecular biology is to understand regulatory relationships between the cellular components. Most of the methods developed to extract gene regulatory relationship from time-delayed gene expression data are not sensitive to non-linearity and non-monotonicity of the cellular system. Here we present four various time-lagged correlation methods including Pearson, Spearman, Kendall and distance correlation and an information theoretic measure (Mutual Information). We propose a method to limit potential regulators while introducing a new dynamic threshold. The SOS DNA Repair of E. coli dataset is used for simulation. The methods are implemented in R Programming language, and the results show the performance of the proposed method to reveal the structure of gene regulatory network.