A hybrid of ant colony optimization and flux variability analysis to improve the production of L-phenylalanine and biohydrogen

In silico metabolic engineering has shown many successful results in genome-scale model reconstruction and modification of metabolic network by implementing reaction deletion strategies to improve microbial strain such as production yield and growth rate. While improving the metabolites production, optimization algorithm has been implemented gradually in previous studies to identify the near-optimal sets of reaction knockout to obtain the best results. However, previous works implemented other algorithms that differ than this study which faced with several issues such as premature convergence and able to only produce low production yield because of ineffective algorithm and existence of complex metabolic data. The lack of effective genome models is because of the presence thousands of reactions in the metabolic network caused complex and high dimensional data size that contains competing pathway of non-desirable product. Indeed, the suitable population size and knockout number for this new algorithm have been tested previously. This study proposes an algorithm that is a hybrid of the ant colony optimization algorithm and flux variability analysis (ACOFVA) to predict near-optimal sets of reactions knockout in an effort to improve the growth rates and the production rate of L-phenylalanine and biohydrogen in Saccharomyces cerevisiae and cyanobacteria Synechocystis sp PCC6803 respectively.