Directionality is an inherent property of biochemical networks

Thermodynamic constraints on reactions directions are inherent in the structure of a given biochemical network. However, concrete procedures for determining feasible reaction directions for large-scale metabolic networks are not well established. This work introduces a systematic approach to compute reaction directions, which are constrained by mass balance and thermodynamics, for genome-scale networks. In addition, it is shown that the nonconvex solution space constrained by physicochemical constraints can be approximated by a set of linearized subspaces in which mass and thermodynamic balance are guaranteed. The developed methodology can be used to {\it ab initio} predict reaction directions of genome-scale networks based solely on the network stoichoimetry.