Article Preview
Top1. Introduction
Nowadays, metabolic engineering is commonly used to produce biofuels such as, alkane, biobutanol, biodiesel, bioethanol, and even hydrogen. Particularly, the high demand for ethanol in the industrial market has led to the increasing amount of ethanol production worldwide. Microorganisms are widely used to produce ethanol via traditional method that is chemical synthesis (Alia et al., 2019). Nonetheless, the production is not yet reached the expectation due to the limitation of the traditional method such as, time consuming, costly, and low raw material usage. Thus, the genome-scale model reconstruction is introduced to retrofit microorganism metabolism to meet industrial demand. Genome-scale model reconstructions are made through the manipulation of genes by genetic engineering processes, such as random mutagenesis and screening. In recent years, genetic engineering was coupled with metabolic engineering, which facilitated the in silico simulation of a genome-scale model reconstruction (Simeonidis & Price, 2015).
Escherichia coli (E. coli) is one of the microorganisms that is often designed as a metabolic model for ethanol production through gene knockout strategies (Woodruff et al., 2013). Gene knockout is a genetic technique that causes inactivation of speciðc genes of an organism to gain its specific function (Song et al., 2011). The knockout of genes causes alteration of a protein structure as some genes encode a subunit of a protein. A subset of reactions is deactivated after a protein is altered and thus the production of compounds is affected. Various mathematical and computational tools have been developed to perform in silico gene manipulations to optimize production of specific compounds. Lately, it has been reported that the available in silico models and computational methods have successfully identiðed a set of genes to be knocked out and optimized the yield of desired product.
Burgard et al. developed OptKnock (Burgard et al., 2003) that is a bilevel optimization framework to suggest gene knockout strategies for the overproduction of a target metabolite while preserving the ñux distribution. Other than gene deletion, the internal ñux distributions such as, growth rate or other biological objectives are optimized. OptReg (Pharkya & Maranas, 2006) is an extension of OptKnock that enhances the algorithm to allow up and/or down regulation for gene knockout to reach the bio-production goal. In OptReg, reaction fluxes are denoted as activated or repressed after their fluxes are enforced to be adequately greater or lower to their corresponding steady-state fluxes, respectively. In addition, RobustKnock (Tepper & Shlomi, 2009) is extended from OptKnock to estimate gene knockout strategies that contribute to the overproduction of the targeted metabolites by considering the competing pathways in the network. More robust prediction is obtained by eliminating all competing pathways, which perhaps obstruct the production rate of the metabolites. However, these methods still face limitations such as, inefficiently handling multivariate and multimodal function and lacking of regulatory or kinetic information in the model that leads to unrealistic ñux distributions. Hence, the production rate from the algorithms may not be fully optimized and still below from the cell theoretical maximum.