Simulation-based systems biology
Understanding the dynamics of biological systems
While classical biology has been very successful in determining the components of living organisms and the mechanistic interactions between individual components, a transition to systems biology is required for the predictive understanding of physiological and pathophysiological dynamics. This transition critically involves the construction of dynamical mathematical models for the biological networks on the individual scales and their integration across scales. The efficient simulation and computational analysis of these models require the development of a new methodological framework based on current simulation technology.
Multi-scale simulation of cellular dynamics
The central aspect of the pursued vision is the implementation of multi-scale models for physiological dynamics in living organism. These models describe the biochemistry and cell population dynamics on scales ranging from dynamics of individual molecules and molecular interactions to the interplay of whole organs in physiological processes. They will be integrated into a simulation environment for predicting physiological dynamics in an organism subject to environmental perturbations on different scales.
Key ingredient 1: Experimental evidence
- Human cancer cells dying by
programmed cell death after different
time intervals upon a common stimulus.
Peeking into life
Recent decades have seen an explosion in measurement techniques directed towards individual molecules, cells, and organ-level dynamics. Yet, experiments tailored to compare measurement data with multi-scale simulation models are difficult and rarely done.
Linking simulation to reality
Experimental researchers in SimTech provide specific measurement data which is required to build predictive simulation models. In addition, they conduct experiments to evaluate simulation-based hypotheses.
Key ingredient 2: Parameter estimation
- Model parameters are sampled using
Markov Chains starting from a-priori
distributions converging to a-posteriori
distributions.
From data to models
Parameter estimation using experimental data is very important on the way to build predictive models for intracellular networks. This task is complicated by uncertain measurements and sparse datasets, especially for quantitative dynamic models, which lead to ill-posed inverse problems.
Estimates with uncertainty
In SimTech we face these difficulties by applying statistical Bayesian approaches for parameter estimation and model fitting. These approaches are particularly suited since they provide besides point estimates also measures of uncertainties.
Key ingredient 3: Uncertainty analysis
- Iterative refinement of the remaining
uncertainty in the activities of key proteins
for insulin signaling.
Quantifying uncertainty
Models of biological systems include significant uncertainties. These affect the quality of predictions based on the model, and therefore need to be characterized by an uncertainty analysis.
Uncertainty and robustness
SimTech develops algorithms for precise predictions of the effects of uncertainty. These are also applied for computing the robustness of a model, which is the level of uncertainty up to which core predictions remain valid.
Key ingredient 4: Model reduction
- A reduced order model for oocyte
population dynamics has been developed,
with a reduction of complexity by a
factor 1000.
Complexity reduction
Reducing the complexity of biological models is essential for integrating them into larger systems. An important requirement here is that the same reduction can be reused when parameters or boundary conditions are changed.
Efficient reduced models
The model reduction methods developed in SimTech allow to construct efficient reduced models where parameters from the original model are retained. These methods have been applied to stochastic models of gene regulation networks.
Key ingredient 5: Visualization
- The visualization supports interactive
zooming from the cellular level onto
single proteins.
Visualizing the data
In systems biology, correlations in high-dimensional data play an important role and good visualizations are essential to make them clear. Visualizations allow to gain deeper insights into simulation results by providing means for quantitative and qualitative analysis.
Coupling different scales
Researchers in SimTech focus on the visualization of simulated data on different scales. The visualization provides support for interactive transitions between the different levels of detail.
(Research Area E)
Key ingredient 6: Ethical Monitoring
- Values both, advances in knowledge and
in technical abilities, change our outlook
on human nature and affect our values.
Shaping the Future
Future application scenarios of systems biological research may raise diverse ethical issues. More subtly, ethical values might also be touched by the change of our understanding of human nature effected by biological paradigms. However, an advantage of simulation technology is that the research process itself remains ethically neutral, because no human or animal experimental subjects are required.
Anticipating Ethical Issues
Systems biology will greatly enhance our ability to intervene in natural processes of living organisms. Ethical reflection is required on a) what kind of manipulations will become possible, b) which of these are desirable or acceptable and which not and c) how misuse can effectively prevented.
(Research Area G)
Demonstrator
SimTech's contributions to Systems Biology are demonstrated via a multi-scale physiological model, describing the distribution of a therapeutic anti-tumor drug through the body and the action of the drug on cells within the tumor. Such models are very valuable for simulation-based therapy planning and optimization.
The simulation challenge here is to couple multiple temporal and spatial scales, large heterogeneities and stochastic effects on the cellular level. The demonstrator shows predictive models on the individual scales, with new coupling strategies, simulation, and visualization methods that allow predicting heterogeneous responses of tumors to drug application.
(Research Areas A, B, C, D, E)
Directly the demonstrator of the vision "Systems Biology"
Contributions to the vision "Systems Biology"