System Biology and the Setting of Standards for Life
Hans V. Westerhoff
Department of Molecular Cell Physiology, Faculty of Earth and Life Sciences, Vrije Universität, Amsterdam, The Netherlands
With the increased accessibility of unicellular life both experimentation and calculation, two dangers arise. One is that of an over emphasis on the collecting of large data sets; “big is beautiful”. The other is that of collecting many data sets under various, uncorrelated experimental conditions. The number of genes being known for organisms, one can calculate the number of triple knock outs that can be studied in terms of their transcriptome. The study of triple knock outs is relevant because of the inherent non-linearity of the dynamics of living organisms, which implies that the study of single knock outs should not suffice to understand how living organisms function. In this presentation we shall illustrate how one may use silicon cells, i.e. computer replica of parts of living cells, under specified standard conditions, to guide the discovery machine of systems biology. This should be accomplished by making modulations of varying extent around the physiological state, rather than making triple or quadruple mutants. Computation and experimentation being executed in parallel, this should provide a lengthy yet efficient way of making biochemistry an biophysics pay off on their promises.
Application of Proteomics in Pharmaceutical Industry
Hanno Langen
Head of Proteomics Initiative, Roche Center for Medical Genomics, Hoffmann-La Roche AG, Basel, Switzerland
Proteomics is a set of technologies that allows the systematic analysis of the protein profile of cells, tissues and body fluids, followed by ever increasing IT-capabilities. Our proteomics approach makes use of the removal of abundant proteins, separation techniques like 2d-gel electrophoresis, capable of separating several hundred proteins in one gel, followed by mass spectrometry for the rapid identification of these proteins. As many steps in the process can be automated (robotic gel processing, mass spectrometry), a high sample throughput can be achieved with high identification rates.
We use a variety of different fractionation methods before we apply 2-dimensional gel electrophoresis. One method is the preparation of subcellular fractions using differential centrifugation and density gradients. Another method is classical chromatography such as ion exchange or hydrophobic interaction chromatography. Some examples will be presented showing the application of these technologies in pharmaceutical research.
Integration of Chemical and Genomic Knowledge in KEGG
Minoru Kanehisa
Bioinformatics Center, Institute for Chemical Research, Kyoto University, Kyoto, Japan
A grand challenge problem in the post-genomic era is a complete computer representation of the cell and the organism, which will enable computational prediction of higher-level complexity of cellular processes and organism behaviors from genomic information. Toward this end we have been developing a knowledge based method for network prediction, which is to predict, given a complete set of genes in the genome, the protein interaction networks that are responsible for various cellular processes. KEGG at www.genome.ad.jp/kegg/ is the reference knowledge base integrating current knowledge on molecular interaction networks such as pathways and complexes (PATHWAY database), the information about genes and proteins generated by genome projects (GENES/SSDB/KO databases), and the information about biochemical compounds and reactions (LIGAND database). These three types of databases actually represent three graph objects, called the protein network, the gene universe, and the chemical universe, respectively. We have developed graph-based methods for analyzing the gene universe and the chemical universe, such as examining genomic association (orthologs across species, adjacency in the genome, etc.) and chemical association (structural families, reaction patterns, etc.) for predicting metabolic networks.
Extending Enzyme Classification with Metabolic and Kinetic Data: Some Difficulties to be Resolved
Keith Tipton
Department of Biochemistry, Trinity College, Dublin, Ireland
Classification of enzymes according to the reaction(s) catalysed is a relatively straightforward problem, since it is dealing with more-or-less factual data. However, attempting to add meaning to those data by adding metabolic or kinetic information takes one into the field of parameters rather than absolutes. Thermodynamic data have been assembled for a number of reactions, but the direction in which a reaction is favoured in isolation does not necessarily mean that will be the direction of the reaction in cellular metabolism; there are many metabolic examples of enzyme reactions proceeding in the thermodynamically less favoured direction. Attempts to predict “missing enzymes” from metabolic pathways should also be treated with caution, since there are several cases where such guesses have proven to be wide of the mark. Incorporation of kinetic data requires the definition of standard conditions, which should ideally bear some relevance to the physiological situation in which the enzyme operates. However, not all enzymes operate under the same physiological conditions, and there are, as yet, no universally-accepted standard conditions, or sets of conditions, of temperature, pH ionic strength etc. for the collection of such data.
The HUPO Proteomics Standards Initiative
Rolf Apweiler
EMBL Outstation - European Bioinformatics Institute, Wellcome Trust Genome Campus, Hinxton, Cambridge, United Kingdom
The Proteomics Standards Initiative (PSI) aims to define community standards for data representation in proteomics and to facilitate data comparison, exchange and verification. Progress has been made in the development of common standards for data exchange in the fields of both mass spectrometry and protein-protein interaction. A proteomics-specific extension is being created for the emerging American Society for Tests and Measurements mass spectrometry standard, which will be supported by manufacturers of both hardware and software.
A data model for proteomics experimentation is under development and discussions on a public repository for published proteomics data are underway. The Protein-Protein Interactions group expects to publish the Level 1 PSI data exchange format for protein-protein interactions soon and discussions as to the content of Level 2 have been initiated. We hope that this meeting will facilitate the creation of a similar workgroup to define community standards for data representation of the functional characterisation of enzymes.
Enzyme Data Requirements for Kinetic and Structural Modelling
David Fell
School of Biological & Molecular Sciences, Oxford Brookes University, Oxford, United Kingdom
Three broad classes of metabolic model are structural, control analysis and kinetic (or dynamic) models. These have successively greater data requirements, with structural models needing only a specification of the reaction catalysed by each enzyme in the model, and kinetic models needing a function describing the response of each enzyme’s rate to every metabolite in the model.
The threonine synthesis pathway of E coli will be used as a case study of building a kinetic model. The kinetics of the pathway enzymes have been studied extensively over the past five decades. Even so, when, with Mazat, Chassagnole and colleagues, I started to build a kinetic model of this pathway to predict its in vivo behaviour, we found very little of the collected data was useful for this purpose. The focus of earlier studies had been the elucidation of allosteric and kinetic mechanisms, with limited concern for furnishing an integrated description of the behaviour of each enzyme under in vivo conditions. As a result, we had to remeasure the kinetic properties of the enzymes, and devise novel equations to combine the quantitative effects of substrates, products and effectors. The model was built using the simulation package SCAMP and successfully simulates, without alteration of the experimentally determined parameters, the main features of the time courses of threonine synthesis measured in cell-free extracts. We have used it to predict the flux control coefficients of the pathway enzymes under physiological conditions at different growth rates, which lead to different levels of demand for threonine.
Even in the apparently more straightforward case of constructing a structural model of a metabolic network, it is not easy to obtain accurate and appropriate information about the component reactions from current data sources. This will be illustrated with examples of the problems arising in collating a structural model of the central metabolism of E. coli using information from various on-line enzyme and genomic databases.
Methods for Design of Optimal Experiments with Application to Parameter Estimation in Enzyme Catalytic Processes
Ekaterina Kostina
IWR, University of Heidelberg, Heidelberg, Germany
Estimating model parameters from experimental data is crucial to reliably simulate dynamic processes. In practical applications, however, it often appears that the experiments performed to obtain necessary measurements are expensive, but nevertheless do not guarantee sufficient identifiability. The optimisation of one or more dynamic experiments in order to maximize the accuracy of the results of a parameter estimation subject to cost and other technical inequality constraints leads to very complex non-standard optimal control problems. more...
Broad-Range Metabolite Analysis: Integration into Genomic Programs
Alisdair Fernie
Department of Molecular Physiology, MPI for Molecular Plant Physiology, Golm, Germany
In recent years the focus of experimental biology has shifted from reductionist towards more holistic approaches. This shift has been driven by the explosion of genetic tools that allowed the creation of an unprecidented base of genetic diversity and by the development of technologies allowing the rapid determination of the genetic, transcriptional, proteins and metabolite complements of biological systems. In this talk I will describe our experiences with broad-range metabolite analysis of potato and tomato development over the last few years. I will describe what can be harvested from these experiments as well as describing recent attempts to analyse systems at the level of more than one molecular entity. Finally, a personal view on the perspective for this research field will be presented.
Determination of Enzyme Activities by Mass Spectrometry - Benefits and Limitations
Hartmut Schlüter
Charité, Benjamin Franklin Campus, Humboldt-University, Berlin
In the field of enzyme kinetics spectrophotometric methods are used extensively to monitor product formation during the enzymatic reation. This requires artificial substrates that undergo a change in absorbance at a given wavelength upon turnover. Although this is a simple and effective means for kinetic analysis, the scope of substrates that can be studied is highly restricted. Furthermore, natural substrates for the enzyme generally cannot be assayed in this manner. Therefore radioactive substrates are often preferred because of their identical chemical nature to the natural substrates as well as their sensitivity of detection. However, radiometric assays require the separation of the radioactive products by thin layer chromatography or other chromatographic methods, and subsequent liquid scintillation counting. Optical as well as radiometric methods both share the problem, that there is an ambiguity about the fate of the chemical structure of the substrate after the enzymatic conversion. Therefore wrong positive results cannot be excluded. A different type of detector, that responds universally to all substrates and reaction products would have definite advantages over spectrophotometric systems. Because enzymatic reactions change the chemical structure of the reactants this change is accompanied by a change in the molecular weight in general. Therefore for the detection of enzymatic activities mass spectrometric techniques are rapid, sensitive and reproducible alternatives. The applicability of mass spectrometry (MS) for the determination of enzyme kinetics was demonstrated by a number of investigators in the past eight years. The first application of MS in conjunction with liquid chromatography for the real-time analysis of enzyme kinetics was reported in 1995 [1]. Bothner et al. demonstrated that in cases where introduction of a chromophore drastically changes the fate of the reaction as a result of the structural features of the substrate electrospray MS (ESI-MS) has been found to be especially valuable [2]. Wu et al. reported that ESI-MS could be used as a rapid, sensitive and accurate quantitative assaying tool for inhibitor libraries [3].
Matrix-assisted laser desorption/ionization (MALDI)-MS is generally more robust than ESI-MS towards buffer solutions and more suitable for complex mixture analysis. It is therefore generally better suited for direct screening of enzyme activities requiring only minimal sample pretreatment and was used to elucidate key fluxes in the central metabolism of lysine producing Corynebacterium glutamicum during batch culture [4]. Nevertheless a restriction for MALDI-MS for quantitative analysis of small enzymatic reaction products is the interference of matrix signals with analyte signals in the mass range between 22 and 500 Da. An appropriate selection of the MALDI matrix may help to solve this problem.
Our group developed a method, named “mass spectrometry assisted enzyme screening (MES)”, by which enzyme activities in complex protein fractions can be measured with a mass spectrometer [5, 6]. The analytical procedure is based on covalent immobilization of proteins to beads. By immobilizing proteins, proteolytic degradation is prevented and the removal of those molecules from the protein fraction is achieved, which otherwise would interfere with the mass spectrometric detection of the enzymatic reaction products. The enzymatic activity is determined by incubating the immobilized proteins with a reaction specific probe, followed by the analysis of the reaction mixture with the MALDI-MS after defined incubation times. Locating a signal in the mass spectrum, which fits the molecular mass of the expected reaction product, validates the type of the enzymatic reaction. MS/MS analysis can provide additional structural information of the analyte being monitored. The method is suitable for both, the analysis of the enzymatic conversion of low molecular weight substances as well as large biopolymers. The MES approach enables the highly sensitive and reliable detection of enzymatic activities even in complex protein mixtures and therefore is a suitable tool for the determination of enzymatic activities in body fluids, cells or tissues.
In conclusion the combined use of enzymology with MS is providing much more detailed insights into the qualitative aspects of enzyme-catalyzed reactions. Unlike the use of chromophore-labeled or radioactive substrates, there is little ambiguity as to the identity of the signal being measured. MS-based determinations of enzymatic activities offer excellent accuracy, reproducibility, and is especially well suited for assaying reactions that cannot be followed photometrically. The small sample size, minimal handling requirements, along with the potential for high-throughput, are further benefits of the combined use of enzymology with MS.
- Hsieh FY, Tong X, Wachs T, Ganem B, Henion J. (1995) Kinetic monitoring of enzymatic reactions in real time by quantitative high-performance liquid chromatography-mass spectrometry. Anal. Biochem. 229, 20-25.
- Bothner B, Chavez R, Wei J, Strupp C, Phung Q, Schneemann A, Siuzdak G. Monitoring Enzyme Catalysis with Mass Spectrometry. (2000) J Biol Chem 275: 13455–13459.
- Wu J, Takayama S, Wong CH, Siuzdak G. (1997) Quantitative electrospray mass spectrometry for the rapid assay of enzyme inhibitors. Chem Biol. 4: 653-7.
- Wittmann C, Heinzle E. (2001) Application of MALDI-TOF MS to lysine-producing Corynebacterium glutamicum: a novel approach for metabolic flux analysis. Eur J Biochem. 268: 2441-55.
- Jankowski J, Stephan N, Knobloch M, Fischer S, Schmaltz D, Zidek W, Schlüter H. (2001) Mass-spectrometry-linked screening of protein fractions for enzymatic activities-a tool for functional genomics. Anal Biochem. 290: 324-9.
- Schlüter H, Jankowski J, Rykl J, Thiemann J, Artsis S, Zidek W, Wittmann B, Pohl T. (2003) Detection of protease activities with the mass-spectrometry-assisted-enzyme-screening (MES) system. Anal Bioanal Chem. In press.
Studying Enzyme Kinetics by Progress Curve Analysis
Hermann-Georg Holzhütter
Institut für Biochemie, Charité, Humboldt Universität, Berlin, Germany
Progress curve analysis, i.e. evaluation of the full time course of an enzyme-catalysed reaction, has become a powerful mean of determining kinetic parameter and discriminating among alternative models. The presentaion will outline the fundamentals of this technique an provide examples.
Structural Modelling of Biochemical Systems: What can we Determine with Minimal Enzymological Information
Mark Poolman
School of Biological & Molecular Sciences, Oxford Brookes University, Oxford, United Kingdom
Even with a perfect set of enzymological for a given (biochemical) system, the would-be modeller may still face problems : The perfect set will neccessarily be known to be valid for only a single organism or cell type under a known set of environmental conditions, and the model will have a large number of real-valued parameters. For a large model, the systematic exploration of the parameter space is not feasible, and it is thus difficult to answer questions concerning the behaviour of a typical, as opposed to a specific, instance of system. Furthermore systems may have interseting properties that are independent of the properties of the enzymes of which they are comprised.
In this talk concepts of structural modelling, dependent only upon the stoichiometries of reactions, will be introduced. Applications of these concepts will be illustrated by describing their application to real-world systems, showing that it is possible to solve problems that would be difficult to approach with kinetic modelling alone. We conclude that structural and kinetic modelling provide a complementary pair of methods for the theoretical investigation of biochemical systems, each able to enhance the information generated by the other.
Multifunctional Enzymes and Pathway Modelling
Stefan Schuster
Department of Bioinformatics, University of Jena, Jena, Germany
Systemic approaches have been used for decades in the modelling of metabolic networks. In particular, metabolic pathway analysis has been developed to study the metabolic capabilities of biotechnologically relevant organisms even if little is known about kinetic parameters. We shall show that this analysis is a useful tool for better understanding the complex architecture of metabolism and for determining the metabolic routes on which the molar conversion yield of a substrate-product pair under consideration is maximum. Modelling in biochemistry has reached a qualitatively new level by the availability of on-line databases from which data about a huge number of enzymes and a large variety of metabolic pathways can be obtained very quickly. Besides information about the encoding genes and kinetic parameters, also information about reaction mechanisms of enzymes is highly valuable. However, the establishment and feeding of these databases meets several problems related with the multifunctionality of many enzymes. First, ontological problems arise concerning the different possible levels of description of substance classes. Second, one needs to decide which independent functions of a multifunctional enzyme are to be listed in the database. Third, it is often unclear which of these functions are really performed in a given organism because this depends on substrate availability.
Beyond Mechanism: Enzyme Kinetics in the Context of Systems Biological Modelling
Jan-Hendrik S. Hofmeyr
Dept. of Biochemistry, University of Stellenbosch, Stellenbosch, South Africa
Classical enzyme kinetics, as developed in the 20th century, had as primary objective the elucidation of mechanism of enzyme catalysis. In this context it proved to be an indispensable theoretical and experimental tool. In systems biology, however, the precise mechanism of an enzyme is less important; what is required is a description of the kinetics of enzymes that takes into account the systemic context in which each enzyme finds itself. This does not mean that the knowledge gained from classical kinetic studies of enzymes is unimportant, but rather that there are additional aspects pertinent to an understanding of enzyme kinetics in systemic context that must be taken into account . Enzyme kinetics in the context of systems biological modelling must take into account (i) the kinetic, regulatory and thermodynamic properties of enzyme-catalysed reactions in the context of the behaviour of coupled reaction systems, i.e., systemic function rather than catalytic mechanism, (ii) the reversibility of all reactions, rather than the artificial irreversible conditions of traditional enzyme kinetic studies, and (iii) those properties of an enzyme that allow us to understand its contribution to cellular control and regulation, such as its elasticity coefficients.
In this lecture I shall illustrate the above aspects in a number of ways. The classical Monod-Wyman-Changeux (MWC) mechanism for cooperativity and allosterism will be compared with the reversible Hill-mechanism developed by Hofmeyr and Cornish-Bowden [1]. Of particular interest is the behaviour with respect to allosteric effector. It is not generally appreciated that in the MWC mechanism the allosteric effector behaves like a pure competitive inhibitor with respect to substrate, i.e., the allosteric effect does not saturate as it does with the Hill-mechanism, where the allosteric effect is abolished at saturating substrate concentrations. The importance of this phenomenon will be discussed in the context of a supply-demand system with feedback. Another important matter to be addressed is whether simpler rate equations for multi-substrate and multi-product reactions, such as power-law equations or equations based on generic mechanisms will suffice for systems biological modelling. These matters are of great importance for the planning of future experiments and for the interpretation of kinetic data from the literature and databases.
[1] Hofmeyr, J.-H.S. & Cornish-Bowden, A. (1997) The reversible Hill-equation: how to incorporate cooperative enzymes into metabolic models. Comp. Appl. Biosci. 13, 377-385.
JWS Online Cellular Systems Modeling and the Silicon Cell
Jacky L. Snoep
Department of Biochemistry, University of Stellenbosch, Stellenbosch, South Africa
After the initial hype of sequencing whole genomes has calmed down it is becoming increasingly clear that for a quantitative kinetic approach we cannot extract sufficient information from the DNA sequence alone. At present it is impossible to predict enzyme properties such as a Km value for a substrate from primary sequence data. These macroscopic properties are of vital importance to build kinetic models, which integrate all the enzyme kinetic characteristics to a quantitative description of multi-enzyme systems. To obtain values for these enzymatic properties we still need to get our hands dirty in the lab. Importantly, the kinetic information that is needed to build these kinetic models is different from what is determined traditionally by enzymologists.
Where enzymologists are interested in the enzyme per se, system biologists have a more physiological interest and need systemic kinetic information, i.e. enzyme kinetic characteristics valid under in vivo conditions. An enzymologist will often work under optimal conditions for maximal enzymatic activity and these might be far removed from the actual conditions under which the enzyme is active in the cell. Typical examples are unrealistic pH values but also absence of product in the reaction incubation of the enzymologist. Kinetic information in protein databases is often determined under non-physiological conditions and lacks information for the reverse reaction, therefore being of limited use for modellers.
A modeller would be looking for a set of data determined under conditions mimicking the intracellular environment as much as possible and taking the reversibility of enzymes into account. Needless to say such data sets are hard to find and the most successful modelling projects include a dedicated experimental group that has measured the complete set of enzyme kinetic parameters under a standard set of conditions. The effort that is put in building such, silicon cell type of models is significant, often several years of work is invested in building these models. This is in stark contrast to the effort that is put in reporting and making the models available to the scientific community. Disturbingly often, model descriptions in the literature are either not complete or contain errors.
Currently, no standard way of model description in the literature exists although the SBML format seems to become the de-facto standard for exchanging models. In December 2000 we started with the development of a database of kinetic models that can be interrogated via the internet (http://jjj.biochem.sun.ac.za or www.jjj.bio.vu.nl) via a user-friendly graphical interface without any software requirement on the client site (a web browser and java runtime environment is all that is needed).
The JWS Online Cellular Systems Modelling is now active for several years with many users and has strong collaborations with other Systems Biology projects such as the Silicon Cell (http://www.siliconcell.net/) and the SBW at Caltech (http://www.sbw-sbml.org/index.html). Furthermore we have started collaborating with the European Journal of Biochemistry and Microbiology to mount models that are described in manuscripts submitted to these journals to our database, first on a secure web site for reviewing and after acceptance on the public database site. In my presentation I will go through the process of building a kinetic model on basis of in vitro enzyme kinetics, using yeast glycolysis as an example.
I will demonstrate kinetic models using the JWS website and indicate the next step towards building the silicon cell; connecting different parts of metabolism by linking different kinetic models together.
Controlled Vocabularies and Ontologies in Enzymology
Kirill N. Degtyarenko
EMBL Outstation - European Bioinformatics Institute, Wellcome Trust Genome Campus, Hinxton, Cambridge, United Kingdom
The diversity of objects and concepts in enzymology can be reflected in number of terms needed to describe uniquely the “elementary” biochemical event such as enzymatic reaction. Consider, for example, the overall enzymatic reaction (including direction) taking place under physiological conditions; any other reaction catalysed by the same enzyme observed in vivo or in vitro; the biochemical pathway the reaction is part of; the mechanism of the enzymatic reaction; an enzyme itself; enzyme subunits. All of these are different classes of entities and as such have to be given their own terms and/or identifiers. In fact, the terminology used e.g. in publications or biological databases is often a mixture of terms borrowed from orthogonal classifications. Although the Enzyme Nomenclature should provide the ultimate controlled vocabulary for a biochemist, it suffers the same problem. EC numbers from a strict hierarchy of ISA relationships and the enzymes often require re-classification. More flexible and in the same time more structured approach was pioneered by Gene Ontology Consortium. I am going to discuss the building of Enzyme Ontology as extension of Gene Ontology. Here, the novel logical relationships unique to chemical entities have to be introduced.
Profiles of Molecular Function – Genomic Enzymology
John Andreassi and Thomas S. Leyh
The Albert Einstein College of Medicine, Bronx, New York, United States of America
The goal of the world-wide structural initiative is to create representative structures for each and every structurally distinct protein domain. A natural extension of this endeavor is to expand our understanding of these structures to include descriptions of their molecular functions. With function in-hand, the domain architecture of a protein becomes a hierarchy of functional entities carefully positioned to accomplish a specific molecular task. We are developing a program to define the molecular functions of highly-conserved, family-defining, surface-residues for each protein family. A primary aim of the program is to establish a database that will allow experimentalists to move quickly from a conserved reside, presented in a structure, to a description of the molecular function of that residue that includes a bioinfomatic map that links the residue to all other pertinent information
(see:http://www7.nationalacademies.org/usnc-ubmb/Presentation_by_Tom_Leyh.html).
Resource Sharing and Data Exchange: SBW and SBML
Herbert M. Sauro
Dept. of Computational Systems Biology, Keck Graduate Institute, Claremont, CA, United States of America
Researchers in quantitative systems biology make use of a large number of different software packages for modelling, analysis, visualization and general data manipulation. In this talk, I will describe the Systems Biology Workbench (SBW), a software framework that allows heterogeneous application components -- written in diverse programming languages and running on different platforms -- to communicate and use each others' capabilities via a fast binary encoded-message system. The goal was to create a simple, high performance, open-source software infrastructure which is easy to implement and understand. SBW enables applications (potentially running on separate, distributed computers) to communicate via a simple network protocol. The interfaces to the system are encapsulated in client-side libraries that are provided for different programming languages. Examples which use SBW will be demonstrated, including interfacing simulators (Jarnac), to model builders (JDesigner), to network analysis tools such as METATOOL (developed by Stefan Schuster et al).
Experimental Enzyme Data as presented in BRENDA
Dietmar Schomburg
CUBIC (Cologne University Bioinformatics Centre), Institute of Biochemistry, Köln, Germany
BRENDA represents a comprehensive information system on enzyme and metabolic information, based on primary literature. The database contains data from at least 83 000 different enzymes from 9800 different organisms, classified in approximately 4200 EC numbers. BRENDA includes biochemical and molecular information on classification and nomenclature, reaction and specificity, functional parameters, occurrence, enzyme structure, application, engineering, stability, disease, isolation, and preparation, links, and literature references. The data are extracted and evaluated from approximately 46 000 references, which are linked to PubMed as long as the reference is cited in PUBMED. In the last year BRENDA underwent major changes including a large increase in updating speed with more than 50% of all data updated in 2002 or in the first half of 2003, the development of a new EC-tree browser, a taxonomy-tree browser, a chemical substructure search engine for ligand structure, the development of controlled vocabulary and an ontology for some information fields, and a thesaurus for ligand names. The database is is accessible free of charge for the academic community at http://www.brenda.uni-koeln.de
Analysis of the experimental data stored in BRENDA shows a number of problems that prohibit a systematic comparison and evaluation of experimental protein data. This is caused by the fact that many experimental data are determined in a non-systematic way and that – on the other hand - the existing recommendations on nomenclature are systematically ignored by most authors of biochemical and molecular-biological papers. Examples will be given.
Enhanced Graphical Representation of Metabolic Networks
Detlef Krömker and Jens Barthelmes
Dept. for Computer Sciences, University of Frankfurt/Main, Frankfurt/Main, Germany
The visualizations currently used to represent biochemical pathways are mostly static and predefined diagrams that have a limited range of applications. They mostly serve as a tool for the visual storage of information rather than for the visual analysis of this information. Due to this fact, they contain only a limited set of data which is often not sufficient for a comprehensive understanding of the underlying processes. This talk will show how an enhanced graphical representation of metabolic networks can change the way in which we work with biochemical pathways. Given that appropriate standard conditions and quantitative data about enzymes and reactions exist, and these data are stored digitally in a suitable format, future graphical representations of pathways could be interactive and integrate information about reaction kinetics, regulatory mechanisms, cellular localization, transport mechanisms, genomic and transcriptional data. Examples of current research activities in this area will be shown and discussed.
Are We Talking about the Same Thing?
Richard Cammack
Department of Life Sciences, King's College London, London, United Kingdom
The aim of biochemical nomenclature is to provide a name for each entity such as a metabolite, an enzyme, or a measured quantity. This should not be the same as, or easily confused with, other terms in biochemistry. There is no single system of nomenclature that is applicable to all cases. There are different requirements, depending on how the name and symbol is to be stored and communicated, by written, printed, or spoken word, as a diagram, or as computer – readable data. For human communication, names are rendered more memorable by relating to biological function, structure or evolutionary relationships. In this respect, nomenclature follows classification. For interaction with computers and databases, identifiers should be unique, extendable, searchable, and referred to an authoritative source. An example is the CAS numbers. Systems of nomenclature for chemical compounds have been developed for many years, and there are well-established rules for deriving a name from a chemical structure, and structure from a name. A recent development is the ICHI (IUPAC Chemical Identifier) which generates a unique, open-source name for each compound, which however needs a computer to read it. For enzymes, there is the EC system which classifies on the basis of the reaction catalysed.
The requirements for nomenclature are distinct from those of a dictionary, where the purpose is to document the way in which names are used in the laboratory and in the literature. The criterion for inclusion of a name in a dictionary is that it is used. Informal systems of nomenclature, or jargon, often proliferate and persist even when they are not systematic and even misleading. When proposing systematic nomenclature, timely intervention is important: not too early when the compounds are inadequately understood, and not too late when misleading terms have become embedded in the literature and databases. When proposing any system of terminology much effort should be devoted to ensuring acceptance of within the biochemical community.
Numerical Optimization Methods for Enzyme Networks
Pedro Mendes
Virginia Bioinformatics Institute, Virginia Tech, Blacksburg, VA, United States of America
Construction of mathematical models of enzyme action based on experimental observations require use of numerical optimization methods. Nonlinear regression of in vitro enzyme kinetic data to obtain estimates of kinetic parameters is an example of the application of numerical optimization that has been practiced for some time now. It is also possible, and desirable, to apply this methodology to experimental settings where one measures several enzymes simultaneously (either in vitro or in vivo). But by scaling up enzyme kinetic nonlinear regressions several technical problems become more severe and different numerical methods are needed that can handle such data. Multi-enzyme data have two characteristics that make them hard for traditional gradient-based algorithms: they are likely to contain many local minima, and they contain regions with very low gradient. Methods such as the popular Lenvenberg-Marquardt either fall on one of these local minima or are unable to proceed due to the small gradient. Global optimization algorithms exist that are more robust towards these problems and are then prime candidates for multi-enzyme kinetic regressions. I will present a summary of what has been found about these algorithms in the context of modeling biochemical networks, and will extend the discussion to the importance of pairing data analysis with experimental design. New large-scale efforts to characterize the kinetic properties of enzymes will benefit from development of robust nonlinear optimization algorithms.
