Cellware: A Modeling and Simulation tool for Large Scale Biological Systems

Sandeep Somani1, Chee Meng2, Li Ye, Anand Sairam, Zhu Hao,Mandar Chitre, Pawan Dhar
1ssomani@bii.a-star.edu.sg, Bioinformatics Institute; 2cheemeng@bii.a-star.edu.sg, Bioinformatics Institute

Recent technological advancements in biology have dramatically enhanced our ability to obtain data, both static and dynamic, at the whole-cell scale. Also the mechanisms of individual processes like cell signaling, gene regulation and metabolic pathways are now better understood. This has set the stage for taking a systems approach to biology wherein the biological networks are studied as an integrated system, as opposed to the traditional reductionist approach. In-silico modeling and simulation are central to the practice of Systems Biology. In the past many efforts were made for extending standard modeling theories to increasingly complex systems using efficient software tools for the simulation. It is clear that a single modeling and simulation approach cannot address the requirements of a diverse array of cellular transactions. Consequently there is a need for a software platform to integrate various mathematical approaches and provide a unified modeling and simulation environment. Cellware is being designed and built as a software platform for large scale model building and simulation efforts. Cellware will help rapid evaluation of various simulation algorithms for deterministic and stochastic pathway simulations, and also model complex phenomenon like diffusion and dynamic network evolution. In addition various analysis tools for parameter estimation, sensitivity analysis and bifurcation analysis will also be available through a single user interface. It is clear that the future in-silico models would be complex and large scale precluding possibility of simulating them on standalone machines. Furthermore parameter estimation and optimization algorithms require a model to run iteratively within a range of parameter values and search space. Therefore the complexity and size of the model coupled with parameter estimation and optimization approaches enforces a high computational demand of resources. To overcome this grand challenge Cellware will use Grid technology with Globus as the middleware. We have also developed an algorithm that combines the flavor of stochasticity with the speed of deterministic systems and will be offered as a default modeling approach in Cellware.