When building a model, researchers are faced with many choices, and typically there are no “right” answers. Researchers must choose the scope of the model (what components and processes to include in the model) and the depth of the model (what details to abstract, and what details to model explicitly). Further, to specify the contents of the model, the researcher must convert mounds of “soft” biological data (for example, phenotype and qualitative in vitro observations) into a concrete and exact mathematical framework. This mapping of fuzzy biological data into a concrete and exact mathematical framework can be difficult, and it is not always obvious what assumptions are being made in the process. Because Cellucidate uses a simple graphical representation of the mathematics that underlies a model, this mapping from a researcher’s fuzzy understanding to a exact modeling statement is made easy.

In the model building process, determination of the model parameters (the concentration/abundances of model components and the reaction rate constants) can be a major challenge. Because researchers generally focus on understanding the qualitative map of the components and interactions in a signaling pathway, we often lack knowledge of the parameters that govern the components and interactions.

Models can be evaluated qualitatively using rough estimates for parameters, or more quantitatively after rigorous estimation of parameter values via an optimization process. Typically, parameter optimization is achieved via an automated process of minimizing the discrepancy between model behavior and quantitative system behavior (experimental data). It is important to note that the resulting parameter estimates generally do not represent robust estimates of the actual underlying reaction rate constants, but rather represent one set of values that result in agreement between the model and experimental data. Estimated parameters may vary widely from the true underlying rate constant for the processes, possibly compensating for other “incorrect” parameters in the model, for missing or inaccurate reactions, or for other inaccurate modeling assumptions.

Despite the large amount of work and knowledge that goes into building models of biological models, the value of the model is generally limited beyond the original use or publication. This is because models are generally only presented in a mathematical format, and not presented and justified in a human readable and understandable format. Standardized formats for model representation, such as SBML, facilitate model distribution but focus largely on the computational or mathematical encoding of the model, and not on the justifications for the choices made during model construction. In the absence of justification of the choices that underlie the components, reactions and parameters in a model, researchers cannot evaluate available models sufficiently well to enable their reuse. Thus, researchers tend to build new models from scratch rather than reuse or extend previously published models.

One goal of Cellucidate is to enable the sharing of models along with sufficient underlying documentation to enable model evaluation, reuse and extension. In addition to providing an intuitive integrated environment for representation of both formal model elements (components, reactions, etc) and justification of these elements, Cellucidate presents the formal model elements in the form of simple human readable cartoons to facilitate understanding of the model.