TY - GEN

T1 - Flux measurement selection in metabolic networks

AU - Megchelenbrink, Wout

AU - Huynen, Martijn

AU - Marchiori, Elena

PY - 2011

Y1 - 2011

N2 - Genome-scale metabolic networks can be reconstructed using a constraint-based modeling approach. The stoichiometry of the network and the physiochemical laws still enable organisms to achieve certain objectives -such as biomass composition- through many various pathways. This means that the system is underdetermined and many alternative solutions exist. A known method used to reduce the number of alternative pathways is Flux Balance Analysis (FBA), which tries to optimize a given biological objective function. FBA does not always find a correct solution and for many networks the biological objective function is simply unknown. This leaves researchers no other choice than to measure certain fluxes. In this article we propose a method that combines a sampling approach with a greedy algorithm for finding a subset of k fluxes that, if measured, are expected to reduce as much as possible the solution space towards the 'true' flux distribution. The parameter k is given by the user. Application of the proposed method to a toy example and two real-life metabolic networks indicate its effectiveness. The method achieves significantly more reduction of the solution space than when k fluxes are selected either at random or by a faster simple heuristic procedure. It can be used for guiding the biologists to perform experimental analysis of metabolic networks.

AB - Genome-scale metabolic networks can be reconstructed using a constraint-based modeling approach. The stoichiometry of the network and the physiochemical laws still enable organisms to achieve certain objectives -such as biomass composition- through many various pathways. This means that the system is underdetermined and many alternative solutions exist. A known method used to reduce the number of alternative pathways is Flux Balance Analysis (FBA), which tries to optimize a given biological objective function. FBA does not always find a correct solution and for many networks the biological objective function is simply unknown. This leaves researchers no other choice than to measure certain fluxes. In this article we propose a method that combines a sampling approach with a greedy algorithm for finding a subset of k fluxes that, if measured, are expected to reduce as much as possible the solution space towards the 'true' flux distribution. The parameter k is given by the user. Application of the proposed method to a toy example and two real-life metabolic networks indicate its effectiveness. The method achieves significantly more reduction of the solution space than when k fluxes are selected either at random or by a faster simple heuristic procedure. It can be used for guiding the biologists to perform experimental analysis of metabolic networks.

UR - http://www.scopus.com/inward/record.url?scp=80455173718&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-24855-9_19

DO - 10.1007/978-3-642-24855-9_19

M3 - Conference contribution

AN - SCOPUS:80455173718

SN - 9783642248542

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 214

EP - 224

BT - Pattern Recognition in Bioinformatics - 6th IAPR International Conference, PRIB 2011, Proceedings

T2 - 6th IAPR International Conference on Pattern Recognition in Bioinformatics, PRIB 2011

Y2 - 2 November 2011 through 4 November 2011

ER -