An
EPSRC Emergent Computing Workshop
|
|
How Do Cells `Decide' to Replicate Their DNA?
Sustained mitogenic stimulation is needed for a cell to cross the G1 checkpoint called the 'restriction point' (R-point) after which the autonomous program for DNA replication is set into motion. The R-point is viewed as a dynamical switch that could be triggered by various intracellular signal transduction pathways including the Ras/MAPK cascade. A proposal is made as to the core mechanism behind the robust R-point switch. This mechanism includes nested positive feedback loops involving the retinoblastoma protein, E2F transcription factors, Cyclin-E/Cdk2, and Cdc25A. Furthermore, the sharpness of the R-point switch is enhanced by the coupling of these positive feedback loops with the negative feedback loop involving p27Kip1 and Cdk2. The 'integrative ability' of the coupled phosphorylation-dephosphorylation cycles, a characteristic of a checkpoint, will be discussed and computer simulations of the operation of the R-point will be presented. |
|
Regulation of Intercellular Fluxes by Gap Junctions
Effective control of intercellular fluxes is essential for the integration of any multicellular system into a functioning unit. In many tissues an intercellular link is provided by arrays of aqueous channels known as gap junctions. Gap junction channels are permeable to ions and molecules up to 1kD, and their permeability is regulated by physiologic factors that modulate the channel kinetics, such as voltage, Ca and pH. Regulation of gap junction permeability is therefore one means of controlling fluxes in cell networks. We have developed a simple simulation tool that can be used to explore flux regulation within a given network of cells linked by gap junctions. The model also offers possible insights into the control of flux in metabolic networks, where now enzymes play a similar role to gap junctions. We will compare the merits of three progressively more complex approaches:
|
|
The Many Faces of a Biological Switch
The cellular environment in a quiescent cell is typically considered to be a relatively stable background against which signalling processes carry out their function. Here we examine the 'background' role of inhibitory protein phosphatases as regulators of a signalling feedback loop. The system was modelled using published experimental results. Quasi-steady-state simulations over a wide range of regulator activities show that the system is bistable over a wide physiological range. The stable/metastable states give rise to a classical catastrophic fold. Thus smooth and slow changes in regulator activity can lead to abrupt, locally irreversible state changes. Simulations of the dynamics of regulation reveal a rich repertoire of responses. These include linear as well as switch-like transients, switch-like outputs with a delayed turnoff, and indefinitely sustained activation. These responses in turn can be modulated by sub-threshold input stimuli applied prior to the test stimulus. Many of these phenomena arise from down- and up-regulation of signaling molecules by the kinases in the feedback loop. This illustrates how a single signalling network can embody multiple kinds of responses with rather complex dynamics. 'Housekeeping' enzymes such as phosphatases appear to play a key role in selecting between various modes of operation of such a network. |
|
Mitotic Clock : How Do Telomeres Tick?
Telomeres, which are chromosome ends, are essential for continuous cellular proliferation. They typically consist of hundreds of copies of 5-8bp sequences, which are rich in G bases at the 3' of DNA strands. Telomeres are involved in protection of chromosomes from recombination and end-degrading enzymes. Their length is maintained by an enzyme telomerase, which is a reverse transcriptase containing an RNA molecule complementary to the telomere repeats of the particular species. Telomerase - independent mechanisms of telomere extension also exist however their biological roles remain to be established. Defects in the maintenance of correct telomere length are involved in human diseases e.g. cancer and the premature aging syndrome, Hutchinson-Gilford progeria. It has been shown that level of telomerase activity depends on the cell type. Sperm cells have a high level of telomerase activity and their telomeres are long and appear not to shorten with age. In contrast, somatic cells usually have low telomerase activity and their telomeres have been estimated to shrink by about 15-40 bp per year. This suggests that telomeric sequences are sustained specifically in germ cells while in somatic cells there is a gradual loss of chromosome ends. It has been postulated that gradual loss of chromosome ends could lead to an exit from the cell cycle. This would mean that chromosomes are incompletely replicated in each round of cell division and shorten to the point when their ends are not protected from degradation and recombination. This would result in senescence and could be a molecular model of aging. Computer simulations testing the telomere theory of cell-aging will be described. |
|
Bifurcation Analysis of a Model of the Embryonic Cell Cycle in Xenopus
Fertilized frog eggs undergo periodic oscillations in the activity of "M-phase promoting factor" (MPF), the major enzymatic activity controlling mitotic cycles in frog eggs, early embryos, and cell-free egg extracts. A complex network of biochemical reactions regulates MPF activity. I have analyzed a mathematical model of this regulatory system developed by Novak and Tyson. Bifurcation theory and numerical methods (AUTO), especially the characterization of the codimension-1 and -2 bifurcation sets of the differential equations, lead to important clues about the behavior of the model. I will describe the bifurcation diagram of this system in a parameter space spanned by the rate constants for cyclin synthesis and cyclin degradation. I will then proceed to detail several hypothesises that are experimentally testable by adding exogenously synthesized cyclin mRNA to extracts depleted of all endogenous mRNA. I will end with a more general discussion of modeling: With regards to dimension, is bigger better? How useful is it to model systems in which parameters (kinetic rate constants, etc) are not available from the experimental literature? |
|
Stochastic Models of Cell Signalling
It is widely expected that computer analysis will be necessary before we can fully understand intracellular signalling pathways. But what is the best way to perform this analysis? Conventional approaches using the numerical integration of continuous, deterministic rate equations offer a convenient route to very large systems or those in which molecular details are not important. But as the resolution of experimental techniques increases so the limitations of conventional models become more evident. Difficulties include the combinatorial explosion of large numbers of different species, the importance of spatial location and conformational changes in the communication process, and the instability associated with reactions between small numbers of molecular species. A radically different approach is to represent individual molecular species as a separate software objects and then to apply Monte Carlo methods to predict the performance of the pathway. In such an approach, rate equations are replaced by individual reaction probabilities and the output has a physically-realistic stochastic nature. Techniques are available by which large numbers of related species can be coded in an economical fashion and key concepts, such as signalling complexes and the thermally-driven flipping of protein conformations, can be embodied into the program. Stochastic modelling may be the way forward that allows us to integrate biochemical and thermodynamic data relating to signal complexes into a coherent and manageable account. |
Using DBsolve for Creating and Analyzing Whole Cell Models
|
|
Future of Computational Molecular Biology in the Pharmaceuticals Industry
The pipeline for the discovery and development of new medicines is now clearly defined within the pharmaceutical industry. This presentation will outline the drug discovery process and the forces driving change in the industry. The benefits of molecular-level systems modelling and its emerging role in the discovery and development of new medicines will be discussed. |
|
Human-Competitive Machine Intelligence by Means of Genetic
Programming
Genetic programming is an automatic programming technique that uses the genetic algorithm to create a computer program to solve a problem. Starting with a primordial ooze of thousands of randomly created computer programs, a population of programs is progressively evolved over a series of generations. The evolutionary search uses the Darwinian principle of survival of the fittest and is patterned after naturally occurring operations, including crossover (sexual recombination), mutation, gene duplication, gene deletion, and certain aspects of the developmental process by which embryos grow into fully developed organisms. There are now over two dozen instances where genetic programming has produced a result that is competitive with a human-produced result. When we talk about competitive with human performance, we mean that the machine-created result infringes on a previously patented invention, duplicates the functionality of a previously patented invention, equals or exceeds a result that was accepted as a new scientific result in a peer-reviewed publication, or is otherwise publishable in its own right (i.e., the result is significant independent of the fact that it was produced by an automated method). These human-competitive results come from the fields of automatic design of analog electrical circuits, automatic design of controllers, quantum computing circuits, cellular automata, sorting networks, and computational molecular biology. We are currently investigating applications of genetic programming to metabolic pathways. |
|
Bioinformatics of Microarray Gene Expression Data
DNA microarray technology is one of the most important recent breakthroughs in experimental molecular biology. With evermore laboratories acquiring this high-throughput technology, the amounts of data being generated are growing extremely rapidly, and the informatics necessary for handling and analysing these data are becoming a major bottleneck. Solving these informatics problems requires standards for the annotation of microarray experiments based on agreed ontologies and controlled vocabularies, where possible. Important tasks that need to be undertaken include: (i) Agreement on the essential information that should be reported for a microarray experiment; (ii) Definition of a standardised structured document format that can capture these data and their semantics, and be easily extensible for future-proofing; (iii) Production of a database that can store these documents; (iv) Development of tools that can search documents in a database and utilize the semantic context to allow comparisons and sophisticated queries. The EBI is working on the development of applications to promote and further the development of the informatics and analysis of microarray data and that is integrated with other biological resources in order to better understand the results of gene expression experiments. We are exploring the development of new techniques as well as technology transfer from marketing and telecommunications domains, e.g. application of visualisation, data mining and statistical analysis. Currently, the EBI is assessing the suitability of different data mining algorithms to microarray gene expression data, e.g. neural networks, classification trees, market basket analysis, clustering, classification, etc. As well developing Internet tools that will allow users to browse and query microarray data stored in a database, the EBI is investigating techniques and technologies that will allow direct access to the microarray information in a database over computer networks, for example CORBA and XML. This technology will allow developers at other sites to develop and write their own novel software tools that can perform queries and analyses on microarray data that is pulled over the Internet from the EBI microarray database. |
|
Tools for Visualization of Genetic Networks Structure and Dynamics
Investigation of structure and complex behavior of genetic networks requires the development of tools for their visualization and interactive exploration. We have proposed a method for representation of genetic network as a directed graph [1] and designed several tools for visualization of genetic networks structure and dynamics using Java programming language. The GeneGraph applet enables the visualization of genetic networks structure at the level of gene expression. It represents genes as rectangles and their interactions as arrows. GeneGraph is apropriate for representation of genetic networks consisting of up to 30 genes. However in the case of large genetic networks the complex intersections of arrows, representing gene interactions, make the comprehension of genetic network structure difficult. To display the structure of large genetic networks we developed the automatic network layout system named as NuT. In this system a good display of genetic network structure is achieved through optimization proceedure which minimizes the number of line crossings in the graph. As an optimization proceedure the serial genetic algorithm is used. The architectute of NuT is based on client-server paradigm. The client consists of Java applet and CGI- interface. The Java applet enables visualizations of genetic network layout before and after optimization procedure. The server is implemented in C and runs genetic algorithm. Both Gene_Graph and NuT are applicable as a Web publishing tools. In particular GeneGraph is already used in GeNet and UrchiNet databases for presentation of a structure of genetic networks [2]. At present the simplest and the most computationally effective model system that gives some insight into the overall behavior of large genetic networks is the Boolean network. We constructed the applet NetWork which enables a user to evaluate the network dynamics in frame of Boolean network model [3]. NetWork presents the possibility to work both with a user defined and with a data provider specified genetic networks. The interactivity of Network and its capability of allowing a user to change the genetic network structure online provides a means of using this tool for assessing the validity of current methods for modeling real genetic networks behavior. References:
|
|
Qualitative Analysis of Gene Networks
Biologists are long aware about the biological and dynamical importance of feedback circuits (i.e. closed chains of regulatory interactions) (see, e.g. (see, e.g., Monod & Jacob, 1961). In particular, positive circuits (i.e. circuits involving an even number of negative interactions) have been associated with the generation of alternative regimes of gene expressions (Lewis et al., 1977; Meinhardt, 1978; Thomas, 1978). Since 1990, much progress has been made in the molecular analysis of gene regulation and development. In the process, dozens of examples of direct or indirect positive auto-regulation of key regulatory genes involved in differentiation have been found (e.g., MyoD, Wg). In parallel, the requirement of positive feedback circuits for multistationarity, as well as that of negative circuits for sustained oscillations have been formally proved by several authors (Plahte et al., 1995; Gouzé, 1998; Snoussi, 1998). Consequently, the biological roles of positive and negative feedback circuits have been further clarified. Whereas negative circuits allow the buffering of gene dosage effects, as well as tight control of the expression of key regulatory genes, positive regulatory circuits may constitute developmental switches, allow the occurrence of alternative developmental pathways, and/or encode positional information. More recently, the notion of feedback circuit proved to be very useful to disentangle complex developmental networks into sets of trans-regulatory modules. In this respect, a functional module is formally defined as a set of interconnecting feedback circuits (e.g. genes). Several illustrations will be provided, including analyses of gene networks controlling pattern formation in Arabidopsis thaliana and Drosophila melanogaster (see also Sanchez et al., 1997; Mendoza et al., 1999). References:
|
|
Robustness and Feedback Control in Biological Signal Transduction
Networks
Living organisms face the challenge of achieving high performance in an uncertain environment with uncertain components. In complex, man-made systems this robustness is conferred by feedback control. We have examined the relationship between feedback control and robustness in two biological signal transduction pathways: (1) bacterial chemotaxis and (2) mammalian phototransduction. We have found that these signaling networks employ standard feedback control strategies to achieve highsensitivity and wide dynamic range in a robust fashion. |
Please send any comments or questions regarding the webpages to alistairR
Last Update: 23rd April 2000
© University of Hertfordshire Higher Education Corporation (2000)