Approximately 30 software tools have been developed for rule-based modeling of biomolecular interaction networks. These tools enable new types of modeling studies. They are particularly useful for investigating biomolecular site dynamics: changes in the states of the functional sites of biomolecules, such as site-specific phosphorylation dynamics. With few exceptions, available software for rule-based modeling is still in a primitive state and critical capabilities are simply unavailable. Existing tools do not provide capabilities that are routinely used in ODE modeling, such as fitting, sensitivity analysis and bifurcation analysis. Moreover, simulators that implement the most generally applicable simulation methods (direct methods) are not being actively developed, and these simulators need to be updated to properly handle certain classes of important problems as well as to offer greater efficiency. We propose to create a toolbox of software tools that will advance the field of computational systems biology. We have identified gaps in existing rule-based software capabilities and present a systematic approach to fill them. Our plan for developing more efficient direct simulation tools involves a two-pronged approach: enabling use of available simulators in distributed computing environments and developing new equation-free computational methods that offer the promise of greater efficiency and integration with existing data analysis software packages. In developing this toolbox, we will improve software for rule-based modeling;integrate existing software tools, and developing new tools for sensitivity and bifurcation analysis and data fitting. These tools are needed so that rule-based modelers can leverage data suited for calibrating parameters of rule-based models, including high-throughput proteomic data. These tools are also needed for diagnosing the dependence of predicted model behaviors on uncertain model parameters, designing experiments to reduce uncertainty in parameter estimates, and elucidating bifurcations (points in parameter space at which sharp transitions in behavior occur). We will test and validate these tools by building a model of receptor tyrosine kinase (RTK) signaling and using this model to investigate how site-specific tyrosine phosphorylation depends on properties of RTK tyrosines and their binding partners. This focus on a driving biological question will ensure that our software development activities are directed at useful capabilities. Our experience developing software tools for rule-based modeling, as well as novel methods, uniquely qualifies us to carry out this proposed project.
The goal of this project is to develop specialized software tools for modeling cellular information processing and decision-making. The deliverables have the potential to facilitate interpretation of high-throughput proteomic data and enable new kinds of modeling studies of cellular regulatory systems.