Improving metabolic flux predictions using absolute gene expression data

BMC Systems Biology

Table 2 Comparison of experimental with predicted exometabolome fluxes, at 85% maximal biomass level

	Experiment	Predicted: Gene expression	Predicted: Standard FBA	Predicted: Fitted FBA	Predicted: GIMME	Predicted: iMAT
Ethanol	13.0	16.2	0	0	0	0
CO₂	21.0	20.1	25.0	21.0	16.0	32.2
Glycerol	2.17	0.126	0	0	0	0
Acetate	0.239	0.00911	0	0	0	0
Trehalose	0.0215	0.0220	0	0	0	0
Lactate	0.00609	0.0176	0	0	0	0
R²		0.96	0.54	0.58	0.52	0.28

Fluxes are reported in units of mmoles/hr/g dry weight (DW), and are scaled by measured glucose uptake flux [see additional file – exometabolomics.xls].
Predicted fluxes are given for the gene expression constrained approach, introduced here, and the standard FBA method that relies upon maximisation of biomass. As standard FBA generates a number of feasible solutions, the “best” solution (the one which minimises the taxicab distance between prediction and data) is reported as “Fitted FBA”. Additionally, the existing algorithms GIMME [18] and iMAT [14] are applied, using the same model and gene expression data as was used to generate the gene expression constrained results.

ISSN: 1752-0509