Human Objective function
In the FBA, the choice of objective function has a deep impact on simulation results.
This issue has not been received much attention in the case of human organism during the past decade due
to the complexity of human metabolic models which includes lots of conditions such as normal,
cancerous, tissue-specificity, and so on.
Although in the study of microorganisms, there is agreement in literature that cell growth (biomass reaction)
is an appropriate choice for an objective function, such selection is still challenging for a human
metabolic model.
In human, healthy cells have a slow growth rate whereas considering a biomass
for an objective function seems to be appropriate for cancerous cells. An understanding of cell
growth is contingent on an understanding of what cells are made of? Or on the other hand, what is
the biomass composition? Generally, it consists of five major components which are
protein, DNA, RNA,
lipids, and carbohydrates with an energy requirement in the form of ATP hydrolysis. There is also
another category called the “other” which is usually taken to include ions and vitamins in order
to satisfy the biological situation. However, it is usually difficult to examine the exact values
for the “other” components. Considering the growth medium is another key feature which is needed
to gain a full understanding of human molecular physiology. It determines which metabolites the
model could consume, on the input side. In other words, to which medium components should the
simulated cells have access?
Some efforts have been performed to explore an applicable objective function for human metabolic models
in the past decade. For instance, Gille et al. tested a liver model with 442 different objectives [1],
in some other studies alternative types of objective functions have been proposed [2-4],
some researchers performed a comparative Flux Variability Analysis (FVA) in order to test the precision
of human metabolic models [5], but none seem to appeal to researchers on an intuitive level the way
biomass objective function [6] does. Therefore, studies in recent years have been performed which
tried to develop a biomass objective function for human metabolic models, but due to the complexity
of a human cell cultures, more attention needed during the simulation study. For example, both
Folger et al. [7] and Wang et al. [8] have almost used the same values for the metabolites appears
in the biomass objective function in their models whereas the former work was done in a cancer network
and the latter study was performed in a normal model. Based on recent development for construction of
tissue-specific metabolic models [8-9], there is need to introduce tissue-specific biomass reactions.
There are new attempts for integration of gene expression data into human metabolic network in order
to construct tissue-specific models [8-10]. These developments could assist researchers to
explore biomass objective functions which would be applicable for human tissue-specific metabolic models.
Although no genome-scale study of human metabolism has included an experimentally determined biomass
composition so far, attempting to gain such purpose could be helpful to develop computational
approaches which be able to compute metabolite coefficients of a biomass reaction based on metabolite
compositions derived from experimental evidences.
Here, we have just tried to inform some challenges regarding to human objective function.
It is necessary for begginer scientist in this field to consider such debates about objective function.
It seems a standard schema needed to be considered for future analysis to help scientists improving other studies continuously.
References:
1. Gille, C., et al., HepatoNet1: a comprehensive metabolic reconstruction of the human hepatocyte for the analysis of liver physiology. Mol Syst Biol, 2010. 6: p. 411.
2. Schuetz, R., L. Kuepfer, and U. Sauer, Systematic evaluation of objective functions for predicting intracellular fluxes in Escherichia coli. Mol Syst Biol, 2007. 3: p. 119.
3. Price, N.D., J.L. Reed, and B.O. Palsson, Genome-scale models of microbial cells: evaluating the consequences of constraints. Nat Rev Micro, 2004. 2(11): p. 886-897.
4. Holzhütter, H.-G., The principle of flux minimization and its application to estimate stationary fluxes in metabolic networks. Eur. J. of Biochem., 2004. 271(14): p. 2905-2922.
5. Boele, J., Of systems and cancer. 2014, Amsterdam Institute for Molecules, Medicines and Systems. p. 166.
6. Feist, A.M. and B.O. Palsson, The biomass objective function. Curr Opin Microbiol, 2010. 13(3): p. 344-9.
7. Folger, O., et al., Predicting selective drug targets in cancer through metabolic networks. Mol Syst Biol, 2011. 7: p. 501.
8. Wang, Y., J.A. Eddy, and N.D. Price, Reconstruction of genome-scale metabolic models for 126 human tissues using mCADRE. BMC Syst Biol, 2012. 6(1): p. 153.
9. Agren, R., et al., Reconstruction of Genome-Scale Active Metabolic Networks for 69 Human Cell Types and 16 Cancer Types Using INIT. Plos Computational Biology, 2012. 8(5).
10. Blazier, A.S. and J.A. Papin, Integration of expression data in genome-scale metabolic network reconstructions. Front Physiol, 2012. 3: p. 299.