Kinetic modelling of metabolic pathways

Reposted from biostars

http://bioinformatics.knowledgeblog.org/2011/06/21/kinetic-modelling-of-metabolic-pathways/

Authors

Natalie J Stanford^1,2 , Kieran Smallbone^2,3
1. Integrative Systems Biology Doctoral Training Centre; 2. Manchester Centre for Integrative Systems Biology; and 3. School of Mathematics, University of Manchester, M13 9PL, UK.

Abstract

We show how to build a kinetic model of a metabolic pathway. We provide the example of building a model of glycerol synthesis using Copasi. However, the techniques required remain the same for any pathway, using any software.

Introduction

A mathematical description of a kinetic metabolic model may be given in differential equation form as $latex x^\prime = N v (x, y, p)$ $latex x(0) = x_0$ and this may be used as a guide as to the data required to create and parameterise a kinetic model. First, N is the stoichiometric matrix, which may be derived from the topology of the model. x denotes metabolite concentrations. y denotes boundary metabolites, whose concentrations are not allowed to vary, but do affect the reaction rates. Initial concentrations for both x and y must be defined, though note that only concentrations x will change over time. Finally, v denotes reaction rates; these are dependent on kinetic mechanisms, concentrations x and y and parameters p.

Stoichiometry

The first stage in kinetic modelling is defining the pathway of interest and its boundaries, which are captured mathematically in the stoichiometric matrix. As an example, we shall use a model of glycerol synthesis containing only two reactions [1], though the same procedure can be used for more complex systems. The model’s two reactions are glycerol 3-phosphate dehydrogenase (GPD) – which has reactants DHAP and NADH and products glycerol 3-phosphate (G3P) and NAD – and glycerol 3-phosphatase (GPP), which has reactant G3P and product glycerol (Gly) and phosphate (Phi). We shall show here how to implement the model using Copasi [2], but the techniques we describe will be applicable to any software. To enter the stoichiometric knowledge in Copasi, navigate to
Model > Biochemical > Reactions
Double click on the equation box, which should take you to the reaction interface (see figure 2). In the chemical equation box enter
DHAP + NADH = G3P + NAD
Click commit and navigate back to Model > Biochemical > Reactions

Figure 1: Navigating to the reaction interface

Figure 2: The reaction interface.

Note: You may notice that Copasi automatically defines a rate law. This is fine, and will be addressed later in the tutorial. Double click the new reaction box, and enter the following in the chemical equation box G3P -> Gly + Phi Note: You may have noticed that in reaction 1 we use = to signify the conversion between substrates and products, whilst reaction 2 uses ->. The former is used to represent a reversible chemical reaction. The latter signifies an irreversible reaction. It is important to note that Copasi will not let you assign pre-defined rate laws for reversible reactions that are irreversible and vice versa. However, if you choose to define your own rate law, these constraints can be violated. This will be discussed in more detail when we address entering rate laws. Navigating now to
Model > Biochemical > Species

Figure 3: The species interface

we see that the six metabolic species have already been added to the model. However, we do not want to simulate changes in concentration of the boundary metabolites in our system. To address this, change Simulation Type in DHAP from reactions to fixed, and repeat this for the other boundary metabolites, Gly, NAD, NADH and Phi.

Figure 4: Changing the metabolite boundaries

The stoichiometry matrix N is now found at
Model > Mathematical > Matrices
As expected, it contains only G3P, which is produced by the first reaction and consumed by the second.

Figure 5: The stoichiometric matrix.

Metabolites

In order to build a kinetic model, concentration values must be provided for all metabolites. Typically these will come from metabolomics measurements, or databases such as HMDB [3]. In Table 1, the concentrations for all metabolites in our model are set out. These may be added by accessing the Species menu in Copasi, and entering the concentration as the initial concentration (see figure 6).

Figure 6: Setting the initial concentration.

ADP, ATP and F16BP are included in the model as modifiers of GPD (Reaction 1). This is represented in Copasi by changing the reaction’s definition to
DHAP + NADH = G3P + NAD; ADP ATP F16BP

metabolite	concentration (mM)
ADP	2.17
ATP	2.37
DHAP	0.59
F16BP	6.01
G3P	0
Gly	15.1
NAD	1.45
NADH	1.87
Phi	1

Table 1: Metabolite concentrations used in the model. It is important to ensure that consistent units are used throughout the model. Navigating to Model, we change Time, Quantity Unit and Volume Unit to min, mmol and l, respectively. In this way, all concentrations are measured in units of mmol/l = mM, and all fluxes in units mM/min.

Figure 7: Model unit interface.

Rate equations

Kinetic rate laws v may be derived from knowledge of the mechanism underlying the enzymatic process. In many cases these are not known, and we must instead resort to approximations such as linlog [4, 5], power law [6, 7] or convenience kinetics [8]. For our model, the rate of GPD is known to be
$latex v_1 = \frac{\frac{Vf1}{K1nadh \cdot K1dhap} \left(NADH \cdot DHAP – \frac{NAD \cdot G3P}{Keq1} \right)} { \left(1 + \frac{F16BP}{K1f16bp} + \frac{ATP}{K1atp} + \frac{ADP}{K1adp} \right) \left( 1 + \frac{NADH}{K1nadh} + \frac{NAD}{K1nad} \right) \left(1+ \frac{DHAP}{K1dhap} + \frac{G3P}{K1g3p} \right)}$
whilst the rate of GPP is given by
$latex v_2 = \frac{\frac{V2 \cdot G3P}{K2g3p}} {\left( 1+ \frac{G3P}{K2g3p} \right) \left( 1+ \frac{Phi}{K2phi} \right)}$
Copasi has a wide variety of inbuilt kinetic laws that may be accessed via the Functions menu. However, the forms above must be entered manually; this is achieved by returning to the Reactions menu and selecting GPD (Reaction 1). Click New Rate Law and enter the following
Vf1/(K1nadh*K1dhap)*(NADH*DHAP-NAD*G3P/Keq1)/((1+F16BP/K1f16bp+ATP/K1atp+ADP/K1adp)*(1+NADH/K1nadh+NAD/K1nad)*(1+DHAP/K1dhap+G3P/K1g3p))
Once the equation has been entered you then need to associate each identifier as a Substrate (NADH, DHAP), Product (NAD, G3P), Modifier (F16BP, ATP, ADP) or Parameter(all other identifiers). This can be seen in figure 8

Figure 8: Entering a new rate law.

Repeat for GPP (Reaction 2)
V2*G3P/K2g3p/((1+G3P/K2g3p)*(1+Phi/K2phi))
once again identifying Substrate (G3P), Product (Gly, Phi) or Parameter (all other identifiers). Returning to the Reactionsmenu, these formulae need to be selected again under Rate Law for each reaction. Where there are multiple substrates, products or modifiers within a reaction it is important to check that the correct identifier is associated with the correct metabolite. These are circled in figure 9

Figure 9: Selecting the rate law.

Kinetic parameters

The kinetic formulae above lead to a range of kinetic parameters. Ideally, these would be measured using enzymatic assay, When this is not possible, databases such as Brenda [9] or Sabio-RK [10] could be used. In Table 2, the parameter values in our model are set out. These can be modified in the same interface highlighted in figure 9.

parameter	value	units
V1	147	mM/min
Keq	10000	1
K1adp	2	mM
K1atp	0.73	mM
K1dhap	0.54	mM
K1f16bp	4.8	mM
K1g3p	1.2	mM
K1nad	0.93	mM
K1nadh	0.023	mM
V2	53	mM/min
K2g3p	3.5	mM
K2phi	1	mM

Table 2: Parameter values used in the model.

Analysis

Having built a model of the system, we would like to use mathematical tools to further our understanding of its behaviour. Copasi provides a number of tools to carry out these tasks under the Tasks menu. In the original paper [1], the authors investigate the response of flux through the pathway to changes in boundary metabolite concentrations. This analysis may be performed running the Sensitivities task, with Function: Concentration Fluxes of Reactions and Variable: Initial Concentrations. Its results are presented in Table 3.

metabolite	response
ADP	-0.141
ATP	-0.420
DHAP	0.475
F16BP	-0.162
G3P	0
Gly	0
NAD	-0.0159
NADH	0.0260
Phi	-0.0741

Table 3: Scaled response coeffcients of flux to metabolite concentrations.

Systems biology standards

Describing mathematical models as above is unwieldy and error-prone, and naturally leads to difficulties in reproduction of results. Thus researchers have developed SBML (the Systems Biology Markup Language [11]). A computer-readable format for representing models of biological processes. Exchanging our models in SBML format (rather than e.g. Copasi’s own xml format) not only allows us to describe our model in unambiguous terms. It also gives us access to over 200 software tools specialising in many elements from data integration to visualisation [12]. SBML can be combined with MIRIAM [13] to annotate the entities of those models; for example, by marking-up the molecule G3P as CHEBI:15978 allows its unambiguous identification and automatically links to many additional sources of information. This may be achieved in Copasi under the Species submenu Annotation. BioModels.net [14] is a modelling repository containing hundreds of models marked up with the above standards. If you are having trouble reproducing the glycerol model described above, it is available from the repository at BIOMD0000000076.

References

1. G. Cronwright and J. Rohwer, “Metabolic control analysis of glycerol synthesis in Saccharomyces cerevisiae,” Applied and Environmental Microbiology, vol. 68, pp. 4448–4456, 2002.
2. S. Hoops, S. Sahle, R. Gauges, C. Lee, J. Pahle, N. Simus, M. Singhal, L. Xu, P. Mendes, and U. Kummer, “COPASI – a COmplex PAthway SImulator,” Bioinformatics, vol. 22, pp. 3067–74, Dec. 2006.
3. D. S. Wishart, D. Tzur, C. Knox, R. Eisner, A. C. Guo, N. Young, D. Cheng, K. Jewell, D. Arndt, S. Sawhney, C. Fung, L. Nikolai, M. Lewis, M.A. Coutouly, I. Forsythe, P. Tang, S. Shrivastava, K. Jeroncic, P. Stothard, G. Amegbey, D. Block, D. D. Hau, J. Wagner, J. Miniaci, M. Clements, M. Gebremedhin, N. Guo, Y. Zhang, G. E. Duggan, G. D. Macinnis, A. M. Weljie, R. Dowlatabadi, F. Bamforth, D. Clive, R. Greiner, L. Li, T. Marrie, B. D. Sykes, H. J. Vogel, and L. Querengesser, “HMDB: the Human Metabolome Database,” Nucleic Acids Res, vol. 35, pp. D521–6, Jan. 2007.
4. D. Visser and J. J. Heijnen, “Dynamic simulation and metabolic re-design of a branched pathway using linlog kinetics,” Metabolic engineering, vol. 5, pp. 164–76, July 2003.
5. K. Smallbone, E. Simeonidis, D. S. Broomhead, and D. B. Kell, “Something from nothing: bridging the gap between constraint-based and kinetic modelling.,” The FEBS journal, vol. 274, pp. 5576–85, Nov. 2007.
6. M. A. Savageau, Biochemical Systems Analysis: A Study of Function and Design in Molecular Biology. Addison-Wesley Pub (Sd), 1976.
7. J. J. Heijnen, “Approximative kinetic formats used in metabolic network modeling,” Biotechnology and bioengineering, vol. 91, pp. 534–45, Sept. 2005.
8. W. Liebermeister and E. Klipp, “Bringing metabolic networks to life: convenience rate law and thermodynamic constraints,” Theor Biol Med Model, vol. 3, p. 41, 2006.
9. M. Scheer, A. Grote, A. Chang, I. Schomburg, C. Munaretto, M. Rother, C. Sohngen, M. Stelzer, J. Thiele, and D. Schomburg, “BRENDA, the enzyme information system in 2011,” Nucleic Acids Research, vol. 39, pp. D670–D676, Nov. 2010.
10. U. Leser, F. Naumann, B. Eckman, U. Wittig, M. Golebiewski, R. Kania, O. Krebs, S. Mir, A. Weidemann, S. Anstein, J. Saric, and I. Rojas, Data Integration in the Life Sciences, vol. 4075 of Lecture Notes in Computer Science. Berlin, Heidelberg: Springer Berlin Heidelberg, 2006.
11. M. Hucka, A. Finney, H. M. Sauro, H. Bolouri, J. C. Doyle, H. Kitano, A. P. Arkin, B. J. Bornstein, D. Bray, A. Cornish-Bowden, A. A. Cuellar, S. Dronov, E. D. Gilles, M. Ginkel, V. Gor, I. I. Goryanin, W. J. Hedley, T. C. Hodgman, J.-H. Hofmeyr, P. J. Hunter, N. S. Juty, J. L. Kasberger, A. Kremling, U. Kummer, N. Le Nov`re, L. M.e Loew, D. Lucio, P. Mendes, E. Minch, E. D. Mjolsness, Y. Nakayama, M. R. Nelson, P. F. Nielsen, T. Sakurada, J. C. Schaff, B. E. Shapiro, T. S. Shimizu, H. D. Spence, J. Stelling, K. Takahashi, M. Tomita, J. Wagner, and J. Wang, “The systems biology markup language (SBML): a medium for representation and exchange of biochemical network models,” Bioinformatics, vol. 19, pp. 524–531, Mar. 2003.
12. SBML software guide
13. N. Le Novère, A. Finney, M. Hucka, U. S. Bhalla, F. Campagne, J. Collado-Vides, E. J. Crampin, M. Halstead, E. Klipp, P. Mendes, P. Nielsen, H. Sauro, B. Shapiro, J. L. Snoep, H. D. Spence, and B. L. Wanner, “Minimum information requested in the annotation of biochemical models (MIRIAM),” Nature Biotechnology, vol. 23, pp. 1509–15, Dec. 2005.
14. “BioModels database – A database of annotated published models”

Tags: copasi, metabolic, modelling, simulation

• Citation

  Please cite this article as: Natalie Stanford,Kieran Smallbone (2011) Kinetic modelling of metabolic pathways. The Bioinformatics Knowledgeblog. http://bioinformatics.knowledgeblog.org/2011/06/21/kinetic-modelling-of-metabolic-pathways/

READ MORE...

OMICs tool directory

It is important to keep abreast of the continuous development of NGS and array software and avoid redundancy. OMICtools provides an overview of more than 2200 tools. All tools have been classified and detailed information provided. By making this resource available, we aim to help both bioinformatics experts and other researchers find appropriate tools for their needs. The directory can also be used by life scientists as an educational or quick-reference support.
 It is a free metadatabase for genomic, transcriptomic, proteomic, and metabolomic data analysis. All tools have been classified by omic technologies (NGS, microarray, MS, NMR). OMICtools is expected to serve as a useful didactic resource not only for bioinformaticians, but also for experimental researchers and clinicians.

 http://omictools.com/

READ MORE...

Stein was right in saying Bioinformatics is a tool,not a field...

He gave the ultimatum on the death of this field in 2003 and in Pakistan where not much NG science is taking place,this has been proved. No jobs for those who specalize in bioinformatics alone.They are considered a half-and-a-half.Not accepted into biology as not enough of a biologist and not accepted into CS disciplines as not specialist of that field.
I agree with Stein that "It is a tool and not a scientific discipline. Tools get absorbed into the greater disciplines. There are examples of disciplines defined by a problem domain contrasted with services defined by tools.
Robust scientific disciplines are often defined by a problem domain. For example, a development biologist studies the development of multicellular organisms using what ever tools are at hand. They aren't defined by the tools they use. A pharmacologist studies the interactions of chemicals with physiological properties. Similarly, physicists aren't defined by their tools; they study the nature of matter and energy.On the other hand, services are defined by tools and they are often time-limited. For example, a microscopist knows how to use microscopes. Now that a microscope is a ubiquitous tool you won't find many specialists in this area. While a pharmacologist has a problem domain, a pharmacist knows how to compound medicines and fill out regulatory paperwork. There are fields that cross over. Stein offered molecular biology as an example of a scientific discipline that has transitioned to a service.
Most of the biologists are bioinformaticians now.It is too central to biology to be left out to specialized bioinformaticians.Colleges and universities should inculcate bioinformatics in the core curriculum.
Bioinformatics was a tool (or set of tools) like microscopy, used to answer complex biological questions. "
Bioinformatics is just one way of studying biology. Whether you think of bioinformatics as High Throughput Biology, Integrative Biology, or Large Data Set Biology, fundamentally Stein argues that bioinformatics is biology.It is not strange when biologists never touch goosh: body parts, liquids.... because they're studying life. Biologists like Ernest Mayer can sit in his office and look at other people's data and develop theories of selection.
Biologists should take programming classes.
Computer has become a central tool for biology like the microscope or the centrifuge. Being able to produce your own software for data analysis should be part of the undergraduate and graduate curriculum. In answer to a related question about computational biology, Stein answered that "it is algorithm development. It is a specialized discipline. I think it's a branch of CS.
Stein has some simple advice for how you make it in bioinformatics:

1. Learn biology. Investigate the problem domain for bioinformatics.
2. Pick a problem that interests you. Don't just follow where you think the hot topic is or what seems to be an easy problem. Consider what you are willing to spend the next decade or two or the rest of your life working on.
3. Know your tools. Don't treat your tools as black boxes. Understand how they work and what their limitations are. With a microscope you don't need to know optics, but you should know something about light paths, magnification, and resolution. Don't be afraid to use non-computer tools. Don't find the problems that fit your tools.
4. Don't be ghettoized. If you expect to be a scientist and to be doing research then don't come in just to perform services to apply your tools to other people's problems. If you want to write software and provide a service, that's great but do so deliberately.
5. Do it because you love it.

 source:http://www.oreillynet.com/pub/a/network/biocon2003/stein.html

Biologists are all bioinformaticians now.

Biologists are all bioinformaticians now.

Biologists are all bioinformaticians now.

READ MORE...

Genome Sequence Could Reveal 'Achilles' Heels' of Important Wheat Disease

Research published in PLoS Genetics on June 9 provides insights into how an important fungal disease is able to evade wheat's defences. The researchers hope that the study, which reveals the fungus' complete genome sequence, will enable them to breed resistant crop plants or improve the use of pesticides.

The genome sequence was produced by an international consortium of researchers including scientists at Rothamsted Research in the UK. The scientists, who were funded by the Biotechnology and Biological Sciences Research Council (BBSRC) and others, are already using the fungus' genome sequence to find ways to control the disease in order to help meet the challenges of ensuring global food security.

The consortium has sequenced the genome of a fungus called Mycosphaerella graminicola which causes leaf blotch disease in wheat. The disease kills cells in the plants leaves leaving large dead blotches which are unable to absorb energy from the sun. This significantly reduces yields and takes a serious toll on wheat crops globally and in the UK; annually a 5-15% reduction in grain yield is incurred in each wheat field solely as a result these infections.

Professor Kim Hammond-Kosack of Rothamsted Research who led the study in the UK said "M. graminicola attacks wheat plants by stealth. There is normally a period of about a week between when a plant first becomes infected and when the characteristic blotches of the disease appear on its leaves. During this time it appears that the plant fails to recognise it has become infected and so is unable to activate its defences to fight back. Studying the fungus' genome will help us to understand how the pathogen is able to go undetected and maybe reveal a chink in its armour that we can exploit."

The genome sequence reveals that M. graminicola has very few genes which produce enzymes able to break-down plant cell walls compared with other fungi which specialise in infecting plants. Plants often use the presence of the sugars and proteins released when a cell is broken down as cue for turning on their immune responses, so the researchers think that the unusually low numbers of genes producing enzymes for breaking down plant cells may be crucial to the fungus' stealth approach.

The team from Rothamsted have already started on work using the genome to look for potential weaknesses in the fungus' defences. Collaborating with researchers at Wageningen University in the Netherlands they have identified a protein in the fungus which is important in keeping it hidden. This research was published recently in Plant Physiology.

Dr Jason Rudd who worked on the project at Rothamsted Research said "We were able to use the information in the genome sequence almost immediately to look for a potential Achilles' heel. We singled out a protein which helps keep the fungus camouflaged and protects it from the plant's defences. When we generated a mutant strain of the fungus which didn't contain the gene for this protein, the infected wheat plants produced strong immune responses and didn't develop the characteristic leaf blotches. Our next step is to use these and similar findings to help farmers combat this disease out in the field in order to reduce wheat losses."

Professor Maurice Moloney, Director and Chief Executive of Rothamsted Research commented "Rothamsted Research is proud to be part of the team that has sequenced and analysed the M. graminicola genome. This fungal pathogen causes one of the most pernicious plant diseases and accounts for significant losses annually in wheat yields throughout the world. This work illustrates the power of sequencing the genomes of plant pathogens in identifying key targets for efficient plant protection."

Professor Douglas Kell, Chief Executive of the Biotechnology and Biological Sciences Research Council, said "Genome sequencing is an important tool in the fight to ensure global food security. Determining the sequence of a destructive crop disease like this is now so quick and affordable that it can be viewed as a research tool rather than a project in itself. This is especially true when the complementary talents of researchers in different countries can be brought to bear. As in this project, this information can be put to immediate use in finding new ways to combat plant disease."

READ MORE...

Paintomics: for visualization of transcriptomics and metabolomics data

The development of the omics technologies such as transcriptomics, proteomics and metabolomics has made possible the realization of systems biology studies where biological systems are interrogated at different levels of biochemical activity (gene expression, protein activity and/or metabolite concentration). An effective approach to the analysis of these complex datasets is the joined visualization of the disparate biomolecular data on the framework of known biological pathways.

Scientists have developed the Paintomics web server as an easy-to-use bioinformatics resource that facilitates the integrated visual analysis of experiments where transcriptomics and metabolomics data have been measured on different conditions for the same samples. Basically, Paintomics takes complete transcriptomics and metabolomics datasets, together with lists of significant gene or metabolite changes, and paints this information on KEGG pathway maps.

Availability: Paintomics is freely available at http://www.paintomics.org.

Contact: aconesa@cipf.es

READ MORE...