Figure 1 .
Steps undertaken in constraining metabolic models with gene expression data. The approach is applicable to genome-scale metabolic model that contain gene-protein-reaction (GPR) relationships. Absolute gene-expression data is mapped to individual reactions following the Boolean logic described in the “Mapping gene expression data to metabolic reactions” section of the Methods. Correlation between this gene-expression data and metabolic fluxes is maximised by following a three step algorithm comprising of: i) maximising the correlation between the initial set of irreversible reactions and the experimental data; ii) performing flux variability to determine additional reactions that must now be unidirectional; iii) repeating this cycle of maximising correlation until no extra irreversible reactions are found through flux variability analysis. The solution predicts exometabolic fluxes that can then be compared to those generated experimentally.
Lee et al. BMC Systems Biology 2012 6:73 doi:10.1186/1752-0509-6-73