A generic Grid interface and execution framework for biomedical applications

Nowadays, there is a growing demand for high computing power and large storage systems in the biomedical domain. Grid computing has recently gained a great deal of interest as an enabling technology towards realization of the e-Science vision, offering a highly flexible and controlled resource-sharing and collaborative environment. However, there is still a lack of user-friendly means to either access the Grid or enable the execution of existing biomedical software into Grid infrastructures. In this work, a generic interface enabling straightforward execution of biomedical applications to Grid infrastructures is presented, with respect to their input/output, software parameterization and attributes for compilation and execution, potentially external files used, etc. This functionality is supported by the definition of a generic XML schema for application description and, accordingly, by the construction of a dynamic graphical user interface based on this schema. Additionally, it provides the means to guide the user towards effective interaction with the Grid, in the entire lifecycle of application execution. The potential and applicability of the proposed approach, as well as the benefits of Grid computing, are illustrated via two existing biomedical applications, namely, the hyperparameter optimization of a Gaussian Support Vector Machine (SVM) classifier applied on human splice site prediction and the simulation of 2D cardiac propagation with the Luo-Rudy I model.