A Framework For Efficiently Identifying Knockout Strategies Leading To Targeted Overproduction Based On Genome-scale Metabolic Networks

In recent years, in silico redesigning methods based on genome-scale metabolic network models have played important roles in metabolic engineering studies and have made some achievements. However, most of these methods are hindered by intractable computing time and can hardly identify comprehensive redesign strategies. In particular, methods that predict knockout strategies leading to overproduction are often limited by allowable knockout number and poor predicted balance between production and growth.Here we propose a new framework named IdealKnock that, identifying knockout targets based on the guidance of overproducing phenotype, is capable of searching knockout strategies efficiently. By testing IdealKnock on the overproduction of a large number of native metabolites as well as on the heterogeneous metabolite lycopene in Escherichia coli model iAF1260, we have shown that the IdealKnock algorithm can successfully break through the limitation of maximum knockout number within 20 minutes and obtain very comprehensive knockout strategies. Besides, IdealKnock also supports for looking for knockout strategies that could keep a very good balance between production and growth. The subsequent combination with the OptKnock framework shows that the obtained knockout strategies by IdealKnock could be simplified efficiently. In addition, by visualizing and analyzing the whole metabolic network, we illustrated the possible biological meanings underlying the suggested strategies in cases of succinate production and lycopene production. Thus, we have shown that the IdealKnock framework can provide very instructive information for metabolic engineering to improve the production of target metabolites.