TY - JOUR

T1 - Dimensional reduction of the master equation for stochastic chemical networks

T2 - The reduced-multiplane method

AU - Barzel, Baruch

AU - Biham, Ofer

AU - Kupferman, Raz

AU - Lipshtat, Azi

AU - Zait, Amir

PY - 2010/8/16

Y1 - 2010/8/16

N2 - Chemical reaction networks which exhibit strong fluctuations are common in microscopic systems in which reactants appear in low copy numbers. The analysis of these networks requires stochastic methods, which come in two forms: direct integration of the master equation and Monte Carlo simulations. The master equation becomes infeasible for large networks because the number of equations increases exponentially with the number of reactive species. Monte Carlo methods, which are more efficient in integrating over the exponentially large phase space, also become impractical due to the large amounts of noisy data that need to be stored and analyzed. The recently introduced multiplane method is an efficient framework for the stochastic analysis of large reaction networks. It is a dimensional reduction method, based on the master equation, which provides a dramatic reduction in the number of equations without compromising the accuracy of the results. The reduction is achieved by breaking the network into a set of maximal fully connected subnetworks (maximal cliques). A separate master equation is written for the reduced probability distribution associated with each clique, with suitable coupling terms between them. This method is highly efficient in the case of sparse networks, in which the maximal cliques tend to be small. However, in dense networks some of the cliques may be rather large and the dimensional reduction is not as effective. Furthermore, the derivation of the multiplane equations from the master equation is tedious and difficult. Here we present the reduced-multiplane method in which the maximal cliques are broken down to the fundamental two-vertex cliques. The number of equations is further reduced, making the method highly efficient even for dense networks. Moreover, the equations take a simpler form, which can be easily constructed using a diagrammatic procedure, for any desired network architecture. It is shown that the method provides accurate results for the population sizes of the reactive species and their reaction rates.

AB - Chemical reaction networks which exhibit strong fluctuations are common in microscopic systems in which reactants appear in low copy numbers. The analysis of these networks requires stochastic methods, which come in two forms: direct integration of the master equation and Monte Carlo simulations. The master equation becomes infeasible for large networks because the number of equations increases exponentially with the number of reactive species. Monte Carlo methods, which are more efficient in integrating over the exponentially large phase space, also become impractical due to the large amounts of noisy data that need to be stored and analyzed. The recently introduced multiplane method is an efficient framework for the stochastic analysis of large reaction networks. It is a dimensional reduction method, based on the master equation, which provides a dramatic reduction in the number of equations without compromising the accuracy of the results. The reduction is achieved by breaking the network into a set of maximal fully connected subnetworks (maximal cliques). A separate master equation is written for the reduced probability distribution associated with each clique, with suitable coupling terms between them. This method is highly efficient in the case of sparse networks, in which the maximal cliques tend to be small. However, in dense networks some of the cliques may be rather large and the dimensional reduction is not as effective. Furthermore, the derivation of the multiplane equations from the master equation is tedious and difficult. Here we present the reduced-multiplane method in which the maximal cliques are broken down to the fundamental two-vertex cliques. The number of equations is further reduced, making the method highly efficient even for dense networks. Moreover, the equations take a simpler form, which can be easily constructed using a diagrammatic procedure, for any desired network architecture. It is shown that the method provides accurate results for the population sizes of the reactive species and their reaction rates.

UR - http://www.scopus.com/inward/record.url?scp=77956110210&partnerID=8YFLogxK

U2 - 10.1103/PhysRevE.82.021117

DO - 10.1103/PhysRevE.82.021117

M3 - ???researchoutput.researchoutputtypes.contributiontojournal.article???

AN - SCOPUS:77956110210

SN - 2470-0045

VL - 82

JO - Physical Review E

JF - Physical Review E

IS - 2

M1 - 021117

ER -