• No results found

Based on the comparison (see Section 5.3), the following findings can be summa- rized for the ten compared programs. The usability of an application is some- what critical from the user’s perspective. For exampleGepasiand BIOCHAMare easy to use, whereas a program such asxCelleratorrequires further programming experience with Mathematica.

Additionally the lack of standards and interfaces between tools becomes ap- parent. For example the support for external file formats, such as SBML, will become more important in the future as the amount of available data increases - “tower of Babel” problem. An example is SmartCell, which usesSBML in com- bination with its own XML file format in such a way that an import of plain

to save a model by one application as an SBML file and reload it with another. This was the case for BIOCHAM and Copasi in both directions.

Then a sufficient documentation and transparency of implementation details are fundamental principles of scientific practice. Here e.g. SmartCell is lacking further information regarding the algorithms it applies. Simulation tools also differ in their ability to utilize external triggers.

Another challenging area is the computational parameter estimation for models. Presently, `a priori information of the parameters may be unavailable and the parameter values are adjusted either with some semi-automatic methods or by manually varying them. The two tools Gepasi and Copasi, provide methods for computational parameter estimation already.

CHAPTER

6

Conclusions

A well structured and extensible platform for systems biology was developed by combining a set of functional components. The already implemented modules for the 4DiCeS application allow for simulation of preliminary applications of biochemical models.

Figure 6.1: Lego drawing of a generalized nucleus. An illustration of the mod- eling of the nucleus by Legoblocks. The blocks can be compared to the 4DiCeS VEs that form a complex structure of cellular compart- ments. In contrast to Lego-blocks however, the VE can be adjusted in size. Thus, the granularity of the system can be adapted to the scientific requirements at hand.

Systems biology poses new challenges for visualization as the data types often contain4Dinformation. The representation of such data is currently unresolved. This work provides an approach to easily handle multi-dimensional data. The presented format could become a standard for 3D modeling environments. In respect thereof this approach works similar to the construction of an object from a box ofLego1 blocks, in whichLegoblocks are exchanged againstVEs (see Figure 6.1). Also the general system’s design can be seen such that each part of the 4DiCeSapplication is exchangeable but will not work alone.

6.1

Design Decisions

With 4DiCeS a modular approach was chosen to allow for the integration of diverse algorithms and file formats. With 4DiCeS models can be composed on different scales and complexity. It allows for the integration of numerous differ- ent algorithms to perform on either single particles, particle concentrations, or distinct VEs. More importantly it is also possible to mix different algorithmic approaches within one simulation space.

In this work the4DiCeSframework was developed for modeling and simulation of signal transduction networks in3D. The implementation of this project includes a modular design of the4DiCeSkernel and various interfaces as described below:

• 4DiCeS-Kernel:

The kernel of the 4DiCeS framework is the core of the project. It is re- sponsible for securely interfacing with other modules. Furthermore it provides a platform for whole cell modeling as a result of its underlying data structure. Here3Dgeometry, reactions, and diffusion are supported. • User Interface:

The user interface facilitates the adaption of tailored user-interaction so- lutions to the 4DiCeS system. Both GUI as well as CLI approaches are

supported.

• Model Parsing Interface:

The input parsing interfaces provide the capability to extend the 4DiCeS

framework to several modeling input description languages. Thus there is the potential to include models from various data repositories.

• Reaction Interface:

The reaction interface allows for the integration of reaction algorithms. These plug-ins provide the actual modular simulation functionality to the

4DiCeS system beside the diffusion interface. • Diffusion Interface:

The diffusion interface assists the combination of both known and new diffusion techniques. It is essential for simulations in3Dto have the ability of translocating particles.

Each interface has its own API, including other programming language exten- sions facilitating the implementation and integration of native code as well as

other language plug-ins. The completeAPIs offer a purpose-specific control over the 4DiCeSplatform.

4DiCeS aims to serve as a universal simulation framework, which can integrate any set of different simulation algorithms, including differential-equation-based models, diffusion-reaction, stochastic algorithms (Gillespie, 1976, 1977, 2001; Gibson and Bruck, 2000) as well as many from CA (Wurthner et al., 2000) to GMA/S-Systems (Hern´andez-Bermejo et al., 2000).

The tool is designed to conduct efficient cellular signaling simulations on a cell model consisting of segments with their specialization on different simulation tasks.

Successful attempts were also made to introduce the visualization environment to a CAVE (see Section 4.2.4) for virtual reality. The user interface from the

4DiCeS package was used for displaying the simulation results onto the four stereo-enabled screens of a CAVE.

Finally the usefulness and utility of this framework was shown for two sample diffusion and reaction applications in comparison to other related tools.