The process flow for evaluating BOSS. A four-step process for evaluating BOSS is illustrated. Specifically, in panel (a), we show how we (1) generate a stoichiometric matrix reconstruction for a biological system and (2) obtain experimental flux measurements for the reactions within the system. Note that when we tested BOSS on simulated noisy flux data in yeast central metabolism, each experimental flux was varied randomly via a normal distribution, with mean and standard deviation equivalent to the mean and standard deviation of the actual experimental fluxes. In panel (b), we show how (3) the reconstruction and experimental flux data are inputted into BOSS and stoichiometric coefficients for the objective reaction are identified through a multiple-restart strategy that utilizes random initial guesses for the coefficients. Additionally, (4) the resultant objectives with a low sum-squared error between the framework-computed flux data and the experimentally-measured flux data are clustered, and the most populous cluster is chosen as representative of the objective function of the system.