Modeling Biological Networks

IV.1 Coordinators
IV.2 Participants
IV.3 Introduction
IV.4 Background and Significance
IV.5 Research Plan
IV.6 Specific Subprojects

• IV.6.iv Subproject 4 Modeling the Regulation and Function of Metabolism
  • IV.6.iv.a Datasources
  • IV.6.iv.b Methods
  • IV.6.iv.c Methodology
    • IV.6.iv.c.1 Cluster analysis
    • IV.6.iv.c.2 Uncovering dynamical constraints using Singular Value Decomposition
  • IV.6.iv.d Research Plan
    • IV.6.iv.d.1 Model construction
    • IV.6.iv.d.2 Model validation
      • IV.6.iv.d.2.i Phenotypic validation
      • IV.6.iv.d.2.ii Molecular validation
      • IV.6.iv.d.2.iii Experimental validation
• [ Other Subprojects ]

IV.7 Connection to Specific Projects 2 (cytoskeleton) and 3 (organogenesis)
IV.8 Timeline

< Previous | Page 23 of 35 | Next >

IV.6.iv.d Research Plan:

Ideally, we would like to construct a model on the skeleton of the metabolic network, including the full reaction kinetics. However, we lack in vivo kinetic data on reaction constants. Also, the computation time is realistic only for smaller modules. However, the relationships the Subprojects reveal between the components allow us to build rather sophisticated semi-quantitative models to predict the constraints limiting the possible behavior of cellular metabolism. Due to the complexity of model building and the need for abundant and reliable data, this Subproject focuses on E. coli. However, the modeling principles will extend to other organisms when data become available.

Our goal is to develop an integrated model that includes the following layers:

1. The stoichiometric matrix describing the topology of E. coli metabolism. The matrix derives from data of very high quality and forms the basic skeleton of our model.
2. The identity of the enzymes associated with each reaction. While both biochemical and genetic identifications exist for most E. coli metabolic reactions (i.e., they possess an identified ORF), microarray analyses will miss those enzymes with only biochemical identifications. For those with defined ORFs, the enzymes that participate in the same modules and pathways are very likely to co-express. Thus enzymatic information will help integrate the regulatory and metabolic networks.
3. The information Subproject 1 reveals about modules in metabolic networks will help us to understand the propagation of perturbations (gene knockouts or changes in external conditions), clarifying the degree to which they affect individual modules.
4. The functional relation between different pathways through the coexpression of enzymes that catalyze participating reactions. As we argued in Subproject 1, in addition to direct reaction-based relations, different reactions and pathways connect through co-regulation. Microarray data partly detect such co-regulation, providing information on the dynamic interaction between different pathways that metabolic topology cannot provide. We will introduce into our models a correlation matrix describing the co-regulation between each pair of enzymes as a separate layer.
5. Clustering based on gene expression data, like clustering based on metabolic network topology, will allow us to see relationships between apparently independent components.

We will continually update the input data for our models from the literature. Since the stochiometric matrix is well established, we expect only minor corrections. However, several ongoing projects may provide dramatic improvements in the quality and quantity of E. coli microarray data.

Our integrated model will offer a comprehensive, systems-level description of E. coli metabolism. The model should provide predictions beyond those provided by looking at metabolic pathways. For example, in a hypothetical experiment we might culture E. coli with an excess of metabolite X. As a result, the biochemical network will assume a new steady state adapted to the new external condition. This change will affect a certain fraction of the network, for example, by a change in the concentrations of those metabolites that react directly with metabolite X. In addition, whole pathways in which X participates may change, and the perturbations may spread to other parts of the metabolic network. Flux balance analysis (Schilling and Palsson, 1998; Schilling et al., 1999; Schilling et al., 2000; Edwards and Palsson, 2000), which we will incorporate into the model, will offer information only on changes that propagate along metabolic pathways. However, as the regulatory network controls the metabolic network, the increase in X will affect the expression pattern of the enzymes involved in the selected pathways. Coexpression analysis at the microarray level of the model will reveal the changes these enzymes cause in other enzymes, which in turn affect metabolic activity. Thus changes in the metabolic network should couple tightly to changes in the regulatory networks. Our integrated model will be able to trace both these effects simultaneously. The model will produce a map that indicates the metabolic pathways and components of the regulatory network that the change in X will most affect, together with the degree (likelihood) of the impact. The model should predict the changes a microarray experiment captures allowing direct comparison to experiment. The same scenario applies to perturbations due to gene removal.