Sequence-based network completion reveals the integrality of missing reactions in metabolic networks

Background: Genome-scale draft metabolic networks are incomplete, even for well studied organisms. Results: Reactions selected by minimizing flux through unlikely reactions resulted in networks of superior quality. Conclusion: Genome-scale models have many network completion solutions but require the addition of unsupported reactions to be functional. Significance: Metabolic networks guide synthetic biology efforts, and the quality of networks determines their predictive power.