The Topology Analysis And Flux Prediction Of Microbial Metabolic Network

Microorganisms thrive in diverse and variable environments through their versatile metabolic capabilities.By converting a variety of different nutrients into a similar essential metabolite pool which provides the energy and biomass for survival and growth under cellular and environmental perturbations.Indeed,this robust functionality of metabolism is manifested in network topology and metabolome dynamics.Topologically,cellular metabolism forms an inhomogeneous scale-scale complex network,which shows high error tolerance—interconnections are always maintained in the face of node(reaction or metabolite)failures and do not require fine-tuning of reaction rate constants or enzymatic capabilities.In metabolome dynamics,both the composition and concentration of metabolomes are conserved across different species.For example,amino acids,particularly glutamate,are the most abundant of all the metabolites detected;the metabolite concentrations are well correlated between E.coli,yeast and mammalian cells.In E.coli cells,the metabolite concentrations in central and even global metabolism remain relatively stable under increasing growth rate and genetic perturbations(single gene deletions),as opposed to the increase in m RNA and protein levels under increasing growth rate.There are a lot of studies which provide some coarse-grained description of the global function of the metabolic systems and detailed analysis of few pathways.However,some fundamental system biology questions,such as,how the stability of intracellular metabolism is depended on the cooperation and coordination between the specific metabolic pathways on genome scale,and how the environment specific adaptability of metabolic function are performed by the structural modularity of metabolic network et.al,are still being explored.In particular,the intracellular metabolisms are the typical of heterogeneous complex systems which are consist of enzymes(reactions),metabolites,cofactors,bioenergy and metal ion etc.,and are involved in almost all the bio-processes of cellular life activities and are dynamically changed with the varying extra-environments.Furthermore,in contrast to the well-built analysis pipeline of the genome-scale of gene expression data,either the completeness of metabolic network model or the collection of experimental metabolomics data,is still in the early stage of the technological development.Hence it is difficult to precisely describe and model the metabolic systems with the mathematical method.Especially,precious studies of metabolic network revealed there are some inconsistent even conflicting results of the nature properties of those networks,for example,the topology analysis of metabolite networks of 47 species indicated the networks are composed of a lot of similar basic structure modules with the organizing principle of the hierarchical scale-free network,as both the node-degree and the clustering coefficient of those networks are followed the power law distribution.While other studies pointed out the observed hierarchical modularity are the result of global connection property of the cofactors instead of the intrinsic nature of metabolite networks.And the biological process analysis of metabolic systems revealed there is a unique core subnetwork of each metabolic network to form a “bowtie” functional modules,while the structure analysis of metabolic network discovered there are at least two dense and highly-connected core subnetworks in each of the metabolic network.Those inconsistency between the metabolic structure and function has seriously lagged the further development of understand of metabolic systems.Profiting from the technological advances,the metabolic network models are becoming increasingly completeness by integrating the new identified enzymes(reactions)into the currency metabolic model and the large-scale metabolomics data is becoming more and more abundant.In this paper,a multi-dimensional systemic biology analysis of the genome-scale of microbial metabolic networks is conducted.And a consistency framework of metabolic network model is constructed based on the new topology properties of the network analysis.Then a new metabolic flux analysis method is conducted on the consistency model of metabolic network by integrating the topological constraints into flux balance analysis,to improve the accuracy of metabolic flux prediction.In detail,the main research of this paper includes:1.The study of topological organizing principle and functional consistency of microbial metabolic network.We have reconstructed the reaction networks of 17 microorganisms by integrating heterogeneous functional components with different properties of the metabolic network into a general structural framework.Here,we reduce network redundancy by correcting the representation of coenzyme factors,removing the repeated connection between the reversible reactions,merging the parallel reactions derived from the different classifications of metabolic enzymes,and re-estimating the current network incompleteness through regression models.The network analysis revealed the observed contradictory topological properties,such as the nonpower-law distribution of node degrees,larger clustering coefficients,and smaller shortest path length,are mainly caused by the indistinguishable representation of complex heterogeneous components and the incompleteness of the current metabolic network models.Interestingly,the positive correlation between the clustering coefficient and the corresponding node degree of each reconstruction network and the power law distribution of node degree of such network revealed that each network has a densely interconnected and large subnetwork,namely core.Decomposition of these heterogeneous reaction networks generates three modules: a highly interconnected and robust core and two linearly sparsely connected peripheral modules corresponding to catabolism and anabolism.Metabolic fluxes generally go from the catabolic module to the core where substantial inter-conversions occur with the flux directions determined by nutrients,and then to the anabolic module.2.Metabolic flux balance analysis based on structural constraintsOur study in last section have revealed the heterogeneous components of metabolic network,especially many metabolic reactions in core subnetwork tend to highly couple with each other to form huge number of reaction cycles and present relative higher kinetic capability.This phenomena suggests these reactions are activated in a cooperative manner.However,the linear programming based metabolic flux prediction methods fail to consider this topology characters.we developed a novel metabolic flux analysis method(named LFBA)for integrating the topological constraints of metabolic network into traditional FBA method to substantially improve the accuracy of predicted metabolic flux without additional omics data,by representing the highly coupled reaction-cycles as the sparse basis vectors of the corresponding stoichiometric matrix.The algorithm test performed on experimental 13 C MFA dataset for multiple growth conditions,by comparing the metabolic flux distribution and growth rate predicted by LFBA with the four traditional metabolic flux balance analysis methods for the three model organisms,the proposed algorithm was better consistent with the experimental measurement than the compared four methods.