LAMOS: A linear algorithm to identify the origin of multiple optimal flux distributions in metabolic networks

Abstract In flux balance analysis, where flux distribution within a cell metabolic network is estimated by optimizing an objective function, there commonly exist multiple optimal flux distributions. Although finding all optimal solutions is possible, their interpretation is a challenge. A new four-phase algorithm (LAMOS) is therefore proposed in this work to efficiently enumerate all of these solutions based on iterative substitution of a current non-basic variable with a basic variable. These basic and non-basic variables are called key reaction pairs that their successive activity or inactivity causes alternate optimal solutions. LAMOS was implemented on E. coli metabolic models and the results proved it as a simple and fast method capable of finding the key reactions as well as reactions participating in the futile cycles. Key reactions were 1–3% of all reactions for the large-scale models and these reactions were identified using only 1% of optimal solutions.

[1]  Kenneth J. Kauffman,et al.  Advances in flux balance analysis. , 2003, Current opinion in biotechnology.

[2]  Motamedian Ehsan,et al.  FLUX DISTRIBUTION IN BACILLUS SUBTILIS: INSPECTION ON PLURALITY OF OPTIMAL SOLUTIONS , 2011 .

[3]  Timo R. Maarleveld,et al.  Interplay between Constraints, Objectives, and Optimality for Genome-Scale Stoichiometric Models , 2015, PLoS Comput. Biol..

[4]  L. Reitzer,et al.  Genetics and Regulation of the Major Enzymes of Alanine Synthesis in Escherichia coli , 2010, Journal of bacteriology.

[5]  Leen Stougie,et al.  Optimal flux spaces of genome-scale stoichiometric models are determined by a few subnetworks , 2012, Scientific Reports.

[6]  Ehsan Motamedian,et al.  A new algorithm to find all alternate optimal flux distributions of a metabolic network , 2015, Comput. Chem. Eng..

[7]  Nan Xu,et al.  Genome-scale reconstruction and in silico analysis of Aspergillus terreus metabolism. , 2013, Molecular bioSystems.

[8]  F. Llaneras,et al.  Stoichiometric modelling of cell metabolism. , 2008, Journal of bioscience and bioengineering.

[9]  I. Grossmann,et al.  Recursive MILP model for finding all the alternate optima in LP models for metabolic networks , 2000 .

[10]  Jörg Stelling,et al.  Large-scale computation of elementary flux modes with bit pattern trees , 2008, Bioinform..

[11]  Hanif D. Sherali,et al.  Linear Programming and Network Flows , 1977 .

[12]  B. Palsson,et al.  Metabolic Flux Balancing: Basic Concepts, Scientific and Practical Use , 1994, Bio/Technology.

[13]  B. Palsson,et al.  Uniform sampling of steady-state flux spaces: means to design experiments and to interpret enzymopathies. , 2004, Biophysical journal.

[14]  Adam M. Feist,et al.  The biomass objective function. , 2010, Current opinion in microbiology.

[15]  R. Mahadevan,et al.  The effects of alternate optimal solutions in constraint-based genome-scale metabolic models. , 2003, Metabolic engineering.

[16]  Markus J. Herrgård,et al.  Integrating high-throughput and computational data elucidates bacterial networks , 2004, Nature.

[17]  L. Buckley,et al.  Cloning of genes that suppress an Escherichia coli K-12 alanine auxotroph when present in multicopy plasmids , 1987, Journal of bacteriology.