Inferring Metabolic States in Uncharacterized Environments Using Gene-Expression Measurements
Environment-specific metabolic models are built in a two-step process. First, gene expression measurements are integrated with the original, environment-independent model. The exploration of alternative flux distributions that maximize the agreement score yields two sets of reactions: zero- and high-frequency reactions. Second, zero-frequency reactions are deleted and the Model Building Algorithm  is used to reduce the network with the constraint that all high-frequency reactions should be able to carry flux.