Möller, Mark;Oleson, Björn E.;Prank, Klaus - 4DiCeS: A Workbench for the Simulation of Intracellular Signalling

ECCB 2002 Poster sorted by: Author | Number

Next | Previous poster (in order of the view you have selected)

Title: 4DiCeS: A Workbench for the Simulation of Intracellular Signalling	P109
Möller, Mark; Oleson, Björn E.; Prank, Klaus mmoeller@techfak.uni-bielefeld.de NRW Graduate School in Bioinformatics and Genome Research

Introduction
Many diseases are based on disorders in intracellular signalling networks, and are not solely based on mutations of single genes. Over the last decades the amount of molecular details on intracellular signal transduction pathways and their interactions has increased tremendously. Biological function arises from the complex interplay between the genome, transcriptome, proteome and the dynamics and location of signalling components. This is one reason why a comprehensive look on the topic of biological signalling is of major relevance.
Due to the high complexity and dynamics of network behaviour (Weng and Bhalla 1999) as well as the large number of components, an intuitive understanding of a system is not possible any more. Thus, mathematical modelling and simulation have become a prerequisite for ana-lysing system dynamics in biology. Simulations of biological signalling allow for an investigation of a system by changing parameters and structure of the networks under study.
A requirement for expanding the use of simulations of biological systems from the computer scientist, mathematician, engineer or physicist to the biologist working on the bench is the development of highly functional, user-friendly simulator systems for the desktop. Due to the increase in computer power in recent years simulations which have high computational demands can now move from supercomputers to the desktop. However, special parallel hardware and software (Cluster, distributed computing) might be still necessary for particular applications.

Our Aims
The objective of our approach is to develop a virtual workbench for biological signalling comprising software modules for:
- automatic generation of mathematical models
- parsing of deterministic ODE model into stochastic models of elementary reactions
- simulation
- visualization
- interfaces to existing databases
- databases for output storage
This tool should help to analyse effects of changes in the parameters and structure of the whole cellular system under study. The simulator will capture the temporal dynamics and location of signalling components such as second messenger, receptors, and subcellular organelles. Based on the real 3D subcelluar architecture of the cell, it shall be possible to calculate changes in the concentration of signalling molecules on any volume element over time.
There exists already a variety of simulators, such as Stochsim (Le Novère and Shmizu, 2001), MCell (De Schutter ,2001), Virtual Cell (Loew and Schaff, 2001), Gepasi (Mendes, 1993) or ECell (Tomita et al., 1999), but they either use deterministic or stochastic approaches and have deficits in simulating the three dimensional subcelluar architecture of a cell.
We think that a hybrid model of deterministic and stochastic components in consideration of the complex subcellular structure will provide more realistic results.
The use of our simulator will be exemplified on the dynamics of signal transduction in primary hepatocytes (liver cells) from rats. This type of cell was chosen due to the major relevance of the liver in metabolism and detoxification of endogenous as well as exogenous sub-stances.
Of course, models are made up of abstractions, but they will be refined more and more in the future due to the increase in knowledge on the molecular level.

Approach
The core of the simulator consists of algorithms for simulating movements of signalling molecules (diffusion) as well as the interactions and reactions of all molecules within subcellular compartments. So far, attempts to model biological signalling are deterministic, based on ordinary (ODE) and partial differential equations (PDE) to represent chemical kinetics. Since ODE and PDE systems used for modelling can not be solved analytically they are integrated to yield numerical results. An alternative approach is the stochastic modelling and simulation of signal transduction, based on the Master equation (McQuarrie, 1967) which is solved numerically by Monte Carlo simulation algorithms.

The deterministic approach regards the time evolution as a continuous, wholly predictable process; whereas the stochastic approach regards the temporal evolution of the biochemical kinetics as a kind of random-walk process.

Realization:
We decided to describe the molecular movements of small numbers of molecule species by a 3D random walk, which has to be fitted to the diffusion coefficient of the substrates in the fluid-gel-like cytoplasm. This way of describing the motion of particles is necessary if we want to simulate the precise path of a signalling molecule from the membrane to the nucleus or other subcellular components.
To save computational time it seemed appropriate to simulate excessive numbers of molecules by shifting them by certain precalculated ratios between subvolumes of a cellular grid

The main problem of precise predictions is that trustful models rely on accurate data. Qualitative and quantitative information gained from experimental work is needed to fill gaps in the understanding of the cellular signalling, which then can be implemented in the simulator. Data mining is another important brick in the construction of this tool. A solution for this problem could be the establishment of direct access from the simulator to worldwide databases of mo-lecular and kinetic information.

The user may describe the pathways he wishes to model by some kind of flow chart, which is then parsed into differential equations as well as a set of elementary reactions suitable for stochastic simulation by Monte Carlo algorithms (Gillespie, 1977; Gibson and Bruck, 2000). Those are then used to calculate the temporal and spatial changes of concentrations within the cell.

Additional information is needed to make predictions of intracellular processes even more precisely. Morphological data of different cell types, the precise location of intracellular compartments and the assembly of membrane substructure can be very helpful to accomplish this goal.

[1] De Schutter, E; Computational Neuroscience: Realistic Modeling for Experimetalists .4 ; CRC Press 2001; ISBN 0-8493-2068-2.
[2] Gibson, M. A.; Bruck, J.; Efficient Exact Stochastic Simulation of Chemical Systems with Many Species and Many Channels; The Journal of Physical Chemistry 2000, 104, 1876-1889.
[3] Gillespie, D. T.; Exact Stochastic Simulation of Coupled Chemical Reactions; The Journal of Physical Chemistry 1977, 81, 2340-2361.
[4] Le Novère, N.; Shimizu, T. S.; StochSim: modelling of stochastic biomolecular processes; Bioinformatics 2001,17, 575-576.
[5] Loew, LM; Schaff, JC; The Virtual Cell: A Software Environment for Computational Cell Biology; Trends in Biotechnology 2001, 19, 401-406.
[6] McQuarrie, D.A.; Stochastic Approach to Chemical Kinetics; J. Appl. Prob. 1967, 4, 413pp.
[7] Mendes, P.; GEPASI: A software package for modelling the dynamics, steady states and control of biochemical and other systems; Comput. Applic. Biosci. 1993, 9, 563-571.
[8] Tomita, M et al.; E-Cell: Software Environment for Whole-Cell Simulation; Bioinformatics 1999, 15, 72-84.
[9] Weng, G; Bhalla U. S.; Complexity in Biological Signaling Systems; Science 1999, 284, 92-96.