Scalable computation of intracellular metabolite concentrations

Quantitative characterization of intracellular metabolite concentrations is central to predictive models cellular functions. Current constraint-based metabolism and macromolecular expression provide genotype-phenotype predictions without accounting for using simplifying assumptions optimality principles. However, principles do not generally apply under stress conditions, rendering these unreliable. Incorporation into or kinetic functions introduces nonlinearities that are computationally challenging handle, limiting their applicability small-scale biological networks. Here, we introduce tractable computational techniques characterize within a modeling framework. This model provides feasible concentration set, which can be nonconvex disconnected. We examine three approaches based on polynomial optimization, random sampling, global optimization. leverage the sparsity algebraic structure underlying biophysical enhance efficiency techniques. then compare performance in two case studies, showing global-optimization formulation exhibits more desirable scaling properties than random-sampling polynomial-optimization formulation, and, thus, promising candidate handling large-scale metabolic