Algorithm for Inference of Gene Regulatory Networks Using Partial Least Squares Regression and Mutual Information

Abstract

Inference of gene regulatory networks refers to identification and localization of functional interactions among genes by application of computational analysis on genomic data. The gene expression data obtained from cDNA microarrays has been used often for such inference. This is an important problem as gene regulatory networks can help in diagnosis and prediction of certain diseases; it is a challenging problem because the data is very high dimensional, noisy, contains very few samples, and time series data is not easily available. The algorithms for inference of gene regulatory networks use the concepts from the fields of information theory, Boolean networks, regression, Bayesian networks, ordinary differential equations, etc. In this paper, we have proposed a method for inference of gene regulatory networks from cancer gene expression data using partial least squares regression to identify important genes, and thereafter, the mutual information is applied to identify the groups of interdependent genes. The results show that the proposed method can identify the biologically relevant networks.