which upon deployment in modeling software can significantly improve the information organization and processing in multiscale cancer systems biology models. We propose the "Cell Cycle Switching Architecture" (CSA) whereby a cellular phenotype can seamlessly hot-switch its cell cycle model during the model simulation, according to its ambient evolutionary pressures. The novel CSA design pattern helps implement vastly different phenotypic cell cycle models and devises their seamless utilization during model simulation.
We also propose the "Embedded Compiler in a Cell" (ECC) architecture. ECC architecture embeds a .NET compiler in the class definition of cellular objects and thus enables the modeler to `stop` an already running simulation, `modify and recompile` the cell cycle model and `resume` the simulation. We have also formalized methods for integrating the legacy multiscale models (implemented in Matlab, C/C++, and Python), by leveraging Microsoft’s .NET Framework. We utilize the .NET wrapper classes as ‘bridges’ for transferring function calls along with their parameters, to the CLR, for execution. We further describe a software design pattern termed “Environmental Field Architecture” (EFA), to help in defining bio-chemical fields for utilization in constructing hybrid multiscale models. EFA represents each nutrient or signaling molecule by a separate environmental ‘layer’. Each layer contains the PDE model and a 2D or 3D matrix representation of the spatial concentration of each environmental molecule. An array of these layers can be used as a representation of a complex environment comprising multiple nutrients and signaling molecules. We envisaged a multi-mode flexibility to add, modify and delete these layers at both design and run time is implemented via separate Graphical User Interface (GUI) and the Software Development Kit (SDK) mode. Alongside, we demonstrate how the n-Tier and Model-View-Controller software design patterns can be employed in the constructing modeling and programming interfaces for multiscale modeling software. We give a practical demonstration of the utility of the above mentioned novel software design patterns, by implementing them in ‘ELECANS’ - a next generation modeling platform for cancer systems biology. ELECANS is equipped with a GUI-based development environment for multiscale modeling along with an SDK such that hierarchically complex biological systems can be conveniently modeled and simulated by using the GUI/SDK combination. Associated software accessories can also help users to perform post-processing of the simulation data for visualization and further analysis.
We followed the software designing and development processes by constructing two multiscale modeling case studies. These case studies utilized the above mentioned software design patterns, and modeled the Warburg Effect and homeostasis in the human colon crypts. In the first case study, we investigate the tumorigenesis provoked by glycolysis and pro-survival autophagy following the mitochondrial permeability transition during cell death. To investigate such mitochondrial dysfunction, we developed a multi-scale model by integrating the dynamic behaviors of essential oncogenic proteins, cells and their microenvironment. We found that (i) the concentration of cellular ATP (adenosine triphosphate) available during the autophagy-related processes is a critical factor in determining tumorigenesis; This thesis describes the design, implementation and applications of an innovative multiagent simulation framework to construct multiscale cancer systems biology models. An enumeration of the novel features implemented in this framework is provided, and is followed up by case studies to demonstrate the usefulness of these features.
Advancements in the biological experimental protocols and the growth of computational infrastructure for the analysis of the resultant experimental data have made it practicable to assemble detailed computational multiscale models. Such computational multiscale models can help cancer biologists to study the spatiotemporal dynamics of complex biological systems and reveal the underlying mechanism of emergent properties. At the heart of multiscale modeling, lies the goal to achieve a comprehensive yet non-redundant computational abstraction of a single cell and its environment, at a minimalistic computational cost.
‘Software Design Patterns’ are defined as software design templates for describing software architecture, in the field of software engineering. Adaptations of such software design patterns to multiscale modeling software can help us to abstract, organize and decode the biological phenomenon from well-defined computational models. The high-throughput era in biology has placed a significant software development requirement on the modeling community. Furthermore, the contemporary software design patterns are not sufficiently developed yet to build representative models of complex biological systems. This lack of software design definition acts as a significant obstacle in the design, development and utilization of novel modeling software.
We have developed an advanced set of software architectural design patterns; (ii) mitochondrial aging rate has a significant influence on this tumorigenic effect, (iii) specific hypoxic and oxidative stresses work cooperatively for tumorigenesis during cell death. We conclude that the cellular mitochondrial status is critical in triggering tumorigenesis during the cell death process, particularly under harsh microenvironments. In the second case study, we assembled a multi-scale model of homeostatic colon crypts. This model was first simulated to achieve steady state homeostatic crypts. The homeostatic crypts were then perturbed to investigate the effects on the crypt’s clonal expansion, in order to obtain a systems-level view of tumorigenesis in Colorectal Cancer (CRC). These investigations included the effects of initial stem cell population, differentiated cell shedding and the symmetric/asymmetric stem cell division ratios on the colon crypt homeostasis. Insights from these investigations unravel the underlying mechanism for maintenance of the colon crypt homeostasis and provide forays into the effects of dysregulation on the crypt morphology.
We conclude that modelers and biologists can utilize the software design patterns described above and its software manifestation (ELECANS), for conveniently constructing their own multiscale modeling platforms. These design patterns provide practicable assistance during the interdisciplinary multiscale software development process. Furthermore, such a software architectural representation for multiscale cancer biology helps to better abstract the underlying biology. This results in the construction of extensible and reusable models as well as modeling software. Further refinements and variations of the reported design patterns can impart an enhanced efficiency to the modeling software design and thus play an important role in the advancement of multiscale modeling in cancer systems biology.