Models as a Service

Mark Walker, Biomedical Engineering


Dissemination of complex biological models is difficult. Just as experimentalists face challenges with the reproducibility of experimental results due to variable reagents, equipment, and technique, modelers often deal with issues such as ill specified parameters, cryptic code, bugs, and disorganized model versions. Galaxy is an open web based platform, designed primarily for genomics research, that provides an unprecedented degree of accessibility, transparency, and reproducibility for data analysis workflows. We introduce the concept of “Models as a Service” and demonstrate how Galaxy is also a suitable platform for running computational simulations. We present a Galaxy tool that runs a computationally demanding of model calcium sparks on Amazon EC2, enabling users to easily experiment with the model in an interactive way. In addition, we have developed a tool that compiles and runs CellML models, demonstrating that Galaxy can be used as a repository for CellML/SBML models. We discuss our experience developing Galaxy tools, its challenges, and share our vision for an online computational model repository.