TopIntroduction
One of the major breakthroughs in the field of molecular biology is the development of a computational branch: famously stated as Bioinformatics. It comprises of two major fields that is molecular biology and information technology. The recent development in this field includes understanding the human genome, new forensic techniques (Gomaa, 2011), as well as the discovery of new medicines of the future (Claverie & Nortredame, 2007). One of the scientific aspects of bioinformatics is that it integrates different fields like machine learning, biology, statistical modeling etc.
Genes being the basic units of heredity in the living organisms plays a major role in the control of cellular processes (Li, Lam, & Shu, 2010).They are the codes of proteins and is transcribed into the corresponding mRNA in the transcription process and then is translated into a protein. With the advancement in the techniques of molecular biology it is possible to measure the gene expression level. Present microarray technology (Schena, Shalon, Davis, Brown, 1995; Shalon, Smith, & Brown, 1996; Ramsay, 1998; Lockhart & Winzeler, 2000) allows measuring the expression levels of thousands of genes in a particular cell at a particular time in a specific condition (Lin, Yeh, Cheng, & Soo, 2007). Hence, they provide information about the gene networks.
Bioinformatics being such an important advancement in the recent development process has one of the most challenging problems: that is to discover relationships and interactions among genes (i.e. discovering gene regulatory networks (Alleyne, 2009)). The network usually represents the regulations among the genes and the basic objective is to extract the expression features, activations and inhibitions from the gene expressions changes of the gene that are present in the microarray data. As the expression of the gene that encodes the regulator, is also regulated by the functional products of some other genes, this forms many complicated regulatory interactions that constitute the structures of underlying Gene Regulatory Networks (GRNs) (Patrick & Keith, 2008).
An intense study of computational biology and the analysis of gene expression networks and metabolic pathways have given rise to various types of GRN (Kauffman, 2007; Savageau, 1996; Arkin,Ross, & McAdams 1998; D’haeseleer, Wen, Fuhrman, & Somogyi, 1999) models. The modeling criterion varies in terms of the details of biochemical interactions, discrete or continuous gene expression level used, deterministic or stochastic approach that has been applied, and so forth (D’haeseleer, Liang, Fuhman, & Somogyi, 2000; Noman & Iba, 2007). These criterions also define how closely the model can represent the genetic interactions. Usually the detailed modeling is very useful for acquiring the precise mechanism in common regulatory pathways. But as we try to simply the network, the complexity of the model increases accordingly. In addition, with the increase in model complexity, the data requirement for learning the model parameters also increases. Despite of various advancement in technology, the microarray technologies (including oligonucleotide arrays and complementary DNA (cDNA) microarrays) were not able to acquire the quality and quantity of data that is required for the accurate reconstruction of genetic networks (Schena, 1999). Therefore, depending on the characteristics of the model used, the availability of the gene expression data, and the level of noise present in the data, the concept of reverse engineering of a genetic network is possible and successful. Thorough review of various integration strategies to support gene regulatory construction was introduced (Chen & VanBuren, 2012) which talked a lot of different distinctive strategies for data integration in gene regulatory network construction.
A massive scale gene construction and testing was done (Gibson, Ficklin, Isaacson, Feltus, & Smith, 2013) using random matrix theory, that showed a huge decrease in network construction time and computational requirement.