The cloud infrastructure is a virtualization platform, that provides access to specific genomic applications and workflows. It is based on OpenNebula to control the underlying hardware infrastructure. Applications are run in virtual machines that are instantiated dynamically following the requirements of the analysis workflow.
Workflows are defined by the use of COMPSs programming model. COMPSs is able to discover implicit parallelism in the pipelines, and hence, execute otherwise serial operations with an optimal use of a parallel environment. COMPSs workflows can be defined using Java, C++, or Python. COMPSs has been adapted to control the virtualization layer, making it transparent to the user, and also allowing to execute the same workflow in a series of environment, from single workstations, to HPC or grid/cloud facilities. Applications where the use of COMPSs would not be desirable can be also executed in their native environment, exploiting already existing parallelism if any.
Complex applications are stored in the system as a collection of pre-packed virtual machines that include the application itself and the necessary software environment. Virtual machines developed here are fully compatible with most common cloud infrastructures.
Not only dynamic virtual machines are supported, but also static web based applications, which can take profit of the HPC facilites if they need to perform heavy back-end calculations.
A list of applications available can be found here.