Stefan Leye

Flexible experimentation in the modeling and simulation framework JAMES II—implications for computational systems biology

Dry-lab experimentation is being increasingly used to complement wet-lab experimentation. However, conducting dry-lab experiments is a challenging endeavor that requires the combination of diverse techniques. JAMES II, a plug-in-based open source modeling and simulation framework, facilitates the exploitation and configuration of these techniques. The different aspects that form an experiment are made explicit to facilitate repeatability and reuse. Each of those influences the performance and the quality of the simulation experiment. Common experimentation pitfalls and current challenges are discussed along the way.

A Discussion on Experimental Model Validation

Model validation is essential in modeling and simulation. It “finalizes” the modeling process, and provides the base for reliable experiments with the model, and thus to gain trustworthy insights of the system under study. Diverse techniques have been developed addressing different needs and are used during different phases in the modeling and simulation life cycle. Experimental model validation depends on the execution of the model. Thus, the peculiarities of the simulation engine might influence the results of experiments and thus have to be taken into account. Execution also underlies some techniques of software validation which can be adopted for experimental model validation. New approaches apply model checking techniques for trace inspections, and emphasize the importance of an explicit description of requirements. All of this implies new requirements for systems intended to support experimental model validation.

Elucidating the Sources of β-Catenin Dynamics in Human Neural Progenitor Cells

Human neural progenitor cells (hNPCs) form a new prospect for replacement therapies in the context of neurodegenerative diseases. The Wnt/-catenin signaling pathway is known to be involved in the differentiation process of hNPCs. RVM cells form a common cell model of hNPCs for in vitro investigation. Previous observations in RVM cells raise the question of whether observed kinetics of the Wnt/-catenin pathway in later differentiation phases are subject to self-induced signaling. However, a concern when investigating RVM cells is that experimental results are possibly biased by the asynchrony of cells w.r.t. the cell cycle. In this paper, we present, based on experimental data, a computational modeling study on the Wnt/-catenin signaling pathway in RVM cell populations asynchronously distributed w.r.t. to their cell cycle phases. Therefore, we derive a stochastic model of the pathway in single cells from the reference model in literature and extend it by means of cell populations and cell cycle asynchrony. Based on this, we show that the impact of the cell cycle asynchrony on wet-lab results that average over cell populations is negligible. We then further extend our model and the thus-obtained simulation results provide additional evidence that self-induced Wnt signaling occurs in RVM cells. We further report on significant stochastic effects that directly result from model parameters provided in literature and contradict experimental observations.

An Efficient and Adaptive Mechanism for Parallel Simulation Replication

Simulation replication is a necessity for all stochastic simulations. Its efficient execution is particularly important when additional techniques are used on top, such as optimization or sensitivity analysis. One way to improve replication efficiency is to ensure that the best configuration of the simulation system is used for execution. A selection of the best configuration is possible when the number of required replications is sufficiently high, even without any prior knowledge on simulator performance or problem instance. We present an adaptive replication mechanism that combines portfolio theory with reinforcement learning: it adapts itself to the given problem instance at runtime and can be restricted to an efficient algorithm portfolio.

A flexible and extensible architecture for experimental model validation

With the rising number and diversity of validation methods, the need for a tool supporting an easy exploitation of those methods emerges. We introduce FAMVal, a validation architecture that supports the seamless integration of different validation techniques. We structure a validation experiment into the tasks specification of requirements, configuration of the model, model execution, observation, analysis, and evaluation. This structuring improves the flexibility of the approach, by facilitating the combination of methods for different tasks. In addition to the overall architecture, basic components and their interactions are presented. The usage of FAMVal is illuminated by several validation experiments with a small chemical model. The architecture has been realized using the plug-in based design of the modeling and simulation framework JAMES II.

One Modelling Formalism & Simulator Is Not Enough! A Perspective for Computational Biology Based on James II

Diverse modelling formalisms are applied in Computational Biology. Some describe the biological system in a continuous manner, others focus on discrete-event systems, or on a combination of continuous and discrete descriptions. Similarly, there are many simulators that support different formalisms and execution types (e.g. sequential, parallel-distributed) of one and the same model. The latter is often done to increase efficiency, sometimes at the cost of accuracy and level of detail. James II has been developed to support different modelling formalisms and different simulators and their combinations. It is based on a plug-in concept which enables developers to integrate spatial and non-spatial modelling formalisms (e.g. stochastic i¾? calculus , Beta binders , Devs , space- i¾?), simulation algorithms (e.g. variants of Gillespies algorithms (including Tau Leaping and Next Subvolume Method ), space- i¾?simulator, parallel Beta binders simulator) and supporting technologies (e.g. partitioning algorithms, data collection mechanisms, data structures, random number generators) into an existing framework. This eases method development and result evaluation in applied modelling and simulation as well as in modelling and simulation research.

Template and Frame Based Experiment Workflows in Modeling and Simulation Software with WORMS

The integration of workflows into modeling and simulation tools promises to provide easier reproduction and provenance of simulation data and its generating process. We present the use of workflow templates and frames realized in WORMS to support and document activities involved in executing simulation experiments. Thereby we make use of functionalities provided by the validation environment FAMVal and the plug-in-based modeling and simulation framework JAMES II. The role of workflows, templates, and frames in modeling and simulation research will be illuminated by a simple simulation study in which the amount of a chemical species in the equilibrium state shall be maximized.

A flexible architecture for performance experiments with the pi-Calculus and its extensions

The π-Calculus is a modeling formalism for concurrent processes. Realized as part of the plug-in based modeling and simulation framework JAMES II, we propose an architecture for π-Calculus-based modeling and simulation, which supports both flexibility and efficiency. Facilitating the design of new π-Calculus-based formalisms and simulators is of particular relevance in the field of computational systems biology, for which many different π-Calculus dialects and simulators have been and still are being developed. Therefore, a flexible representation of π-Calculus models is used, which is illustrated by a mapping from the biochemical variant of the π-Calculus to the representation. Simulation engines are exchangeable and even automatically configurable according to the task at hand. Moreover, we present three different simulator implementations, working on the model representation. Efficiency denotes that our architecture supports the implementation of high-performance simulators. In order to assess efficiency, we perform experiments with these simulators and compare the results to the current cutting edge implementation in the field, the Stochastic Pi Machine.

A flexible architecture for modeling and simulation of diffusional association

Up to now, it is not possible to obtain analytical solutions for complex molecular association processes (e.g. Molecule recognition in Signaling or catalysis). Instead Brownian Dynamics (BD) simulations are commonly used to estimate the rate of diffusional association, e.g. to be later used in mesoscopic simulations. Meanwhile a portfolio of diffusional association (DA) methods have been developed that exploit BD. However, DA methods do not clearly distinguish between modeling, simulation, and experiment settings. This hampers to classify and compare the existing methods with respect to, for instance model assumptions, simulation approximations or specific optimization strategies for steering the computation of trajectories. To address this deficiency we propose FADA (Flexible Architecture for Diffusional Association) - an architecture that allows the flexible definition of the experiment comprising a formal description of the model in SpacePi, different simulators, as well as validation and analysis methods. Based on the NAM (Northrup-Allison-McCammon) method, which forms the basis of many existing DA methods, we illustrate the structure and functioning of FADA. A discussion of future validation experiments illuminates how the FADA can be exploited in order to estimate reaction rates and how validation techniques may be applied to validate additional features of the model.

Composing problem solvers for simulation experimentation: a case study on steady state estimation.

Simulation experiments involve various sub-tasks, e.g., parameter optimization, simulation execution, or output data analysis. Many algorithms can be applied to such tasks, but their performance depends on the given problem. Steady state estimation in systems biology is a typical example for this: several estimators have been proposed, each with its own (dis-)advantages. Experimenters, therefore, must choose from the available options, even though they may not be aware of the consequences. To support those users, we propose a general scheme to aggregate such algorithms to so-called synthetic problem solvers, which exploit algorithm differences to improve overall performance. Our approach subsumes various aggregation mechanisms, supports automatic configuration from training data (e.g., via ensemble learning or portfolio selection), and extends the plugin system of the open source modeling and simulation framework James II. We show the benefits of our approach by applying it to steady state estimation for cell-biological models.