Association rule mining for the identification of activators from gene regulatory network

Abstract

Recent advances in Microarray technologies have encouraged to extract gene regulatory network from microarray data in order to understand the gene regulation (in terms of activators and inhibitors) from time-series gene expression patterns in a cell. The concept of positive and negative co-regulated gene clusters (pncgc)[1] Association Rule Mining is used to analyze the gene expression data that more accurately reflects the co-regulations of genes than the existing methods which are computationally expensive. Experiments were performed with Saccharomyces cerevisiae and Homo Sapiens dataset through which semi co-regulated gene clusters and positive and negative co-regulated gene clusters were extracted. The resulting semi co-regulated gene clusters were used in inferring a gene regulatory network which was compared with large scale regulatory network inferred from modified association rule mining algorithm. The usage of positive and negative co-regulated gene cluster approach of identifying the network outperformed the modified association rule mining [2], especially when analyzing large numbers of genes. © 2011 Springer-Verlag.