A.S. Sidhu & S.K. Dhillon (Eds.): Adv. in Biomedical Infrastructure 2013, SCI 477, pp. 5–14. DOI: 10.1007/978-3-642-37137-0_3 © Springer-Verlag Berlin Heidelberg 2013 Inferring E. coli SOS Response Pathway from Gene Expression Data Using IST-DBN with Time Lag Estimation Lian En Chai, Mohd Saberi Mohamad * , Safaai Deris, Chuii Khim Chong, and Yee Wen Choon Artificial Intelligence and Bioinformatics Research Group, Faculty of Computer Science and Information Systems, Universiti Teknologi Malaysia, Skudai 81310, Johor, Malaysia {lechai2,ckchong2,ywchoon2}@live.utm.my, {saberi,safaai}@utm.my Abstract. Driven to discover the vast information and comprehend the fundamental mechanism of gene regulations, gene regulatory networks (GRNs) inference from gene expression data has gathered the interests of many researchers which is otherwise unfeasible in the past due to technology constraint. The dynamic Bayesian network (DBN) has been widely used to infer GRNs as it is capable of handling time-series gene expression data and feedback loops. However, the frequently occurred missing values in gene expression data, the incapability to deal with transcriptional time lag, and the excessive computation time triggered by the large search space, are attributed to restraint the effectiveness of DBN in inferring GRNs from gene expression data. This paper proposes a DBN-based model (IST-DBN) with missing values imputation, potential regulators selection, and time lag estimation to address these problems. To assess the performance of IST-DBN, we applied the model on the E. coli SOS response pathway time-series expression data. The experimental results showed IST-DBN has higher accuracy and faster computation time in recognising gene-gene relationships when compared with existing DBN-based model and conventional DBN. We also believe that the ensuing networks from IST-DBN are applicable as a common framework for prospective gene intervention study. Keywords: Dynamic Bayesian network, missing values imputation, time-series gene expression data, gene regulatory networks, network inference. 1 Introduction In the post-genomic era, aided by the breakthroughs in technology, researchers have begun to shift the research paradigm from the classical reductionism to the modern holism, wherein biological systems and experimental design are viewed as a whole instead as collections of parts [1]. One of the innovations conceived in such era, the * Corresponding author.