Evaluating noise in the flux data in the S. cerevisiae central metabolic network for the precursor biomass synthesis objective reaction. The results of BOSS when noisy flux data are supplied as input to the framework are illustrated in the context of the precursor biomass synthesis objective reaction. Panel (a) illustrates the sum-squared error between the computed objective reaction and the hypothesized objective reaction of precursor biomass synthesis, when normalized to the magnitude of the precursor biomass synthesis reaction. The mean and standard deviation represented by the error bars for SSE
at each of the percent variances are based on the 20 independent runs at each level of noise (i.e., 20 independent flux distributions with equivalent amounts of noise randomly inserted). Panels (b), (c), and (d) illustrate the hypothesized objective reaction (gray bars) and the BOSS-derived objective reaction (blue dots/lines) when the reaction fluxes in the network are varied by 5, 15, and 25 percent of their actual values, respectively. As the variance in the fluxes increases, the performance of BOSS degrades. Note that, as described previously in the context of Figures 4(c) and 4(d), to avoid biasing results, the hypothesized objective reaction of precursor biomass synthesis was included in the network inputted into BOSS, and the pool of experimental fluxes consisted of all fluxes with the exception of the precursor biomass synthesis reaction flux.