On the computation of the minimum set of reactions for optimal growth in constraint-based models.

Technical advances in sequencing have allowed the reconstruction of metabolic models of multiple microorganisms, which have proven useful in advancing metabolic engineering and drug discovery. Optimization methods have provided a way to accurately predict flux phenotypes of various unicellular organisms and their response to gene knockouts. Despite the broad application of these methods, the role that different biochemical reactions have in providing robustness and flexibility has not been studied extensively. In this work, a method is presented to identify those sets of reactions that are essential for growth and those that are redundant and therefore account for the robustness of metabolism. The problem of computing a minimum set of reactions that can produce optimum growth is formally stated. It is proven that such a problem is NP-complete and a technique to reduce the search space of the problem is proposed. The presented approach is experimentally applied in various genome-scale models. The contribution of this work is to provide insight into the roles that different reactions play in the production of growth and to propose methods that can be directly applied in model curation and analysis.