Open Access Highly Accessed Methodology article

Improving metabolic flux predictions using absolute gene expression data

Dave Lee1, Kieran Smallbone1, Warwick B Dunn1, Ettore Murabito1, Catherine L Winder1, Douglas B Kell12, Pedro Mendes13 and Neil Swainston1*

Author affiliations

1 Manchester Institute of Biotechnology, University of Manchester, 131 Princess Street, Manchester, M1 7DN, UK

2 School of Chemistry, University of Manchester, Manchester, M13 9PL, UK

3 Virginia Bioinformatics Institute, Virginia Tech, Washington St. 0477, Blacksburg, Virginia, 24060, USA

For all author emails, please log on.

Citation and License

BMC Systems Biology 2012, 6:73  doi:10.1186/1752-0509-6-73

Published: 19 June 2012

Abstract

Background

Constraint-based analysis of genome-scale metabolic models typically relies upon maximisation of a cellular objective function such as the rate or efficiency of biomass production. Whilst this assumption may be valid in the case of microorganisms growing under certain conditions, it is likely invalid in general, and especially for multicellular organisms, where cellular objectives differ greatly both between and within cell types. Moreover, for the purposes of biotechnological applications, it is normally the flux to a specific metabolite or product that is of interest rather than the rate of production of biomass per se.

Results

An alternative objective function is presented, that is based upon maximising the correlation between experimentally measured absolute gene expression data and predicted internal reaction fluxes. Using quantitative transcriptomics data acquired from Saccharomyces cerevisiae cultures under two growth conditions, the method outperforms traditional approaches for predicting experimentally measured exometabolic flux that are reliant upon maximisation of the rate of biomass production.

Conclusion

Due to its improved prediction of experimentally measured metabolic fluxes, and of its lack of a requirement for knowledge of the biomass composition of the organism under the conditions of interest, the approach is likely to be of rather general utility. The method has been shown to predict fluxes reliably in single cellular systems. Subsequent work will investigate the method’s ability to generate condition- and tissue-specific flux predictions in multicellular organisms.

Keywords:
Flux balance analysis; Metabolic flux; Metabolic networks; Transcriptomics; RNA-Seq; Exometabolomics