Get Started
(beta release)


Cloud Simulation Service Powered by Flint


Because biophysiological models are becoming progressively larger, high performance computers and simulators such as Flint are increasingly necessary. Flint K3, a cloud-based simulation service, has been developed so that users can process simulation jobs in a high performance computing environment without the cost of high performance hardware. Models must be written in PHML (Physiological Hierarchy Markup Language) or SBML (Systems Biology Markup Language).

K3 is composed of two servers (See figure below). One is an interface server (IFS) that receives job requests from users and manages the jobs. Simulation jobs are sent from the IFS to simulation servers (SS) on the cloud, which are powered by the Flint computation engine. SSs launch computation programs (CPs) at every node assigned by the simulation. CPs perform parallel numerical computations as required by the model, communicating among themselves. When the simulation finishes, CPs terminate automatically.

We have started K3 with “Academic Inter-Cloud'', a bare metal community cloud with modified OpenStack, developed at the National Institute of Informatics (NII). For development and preliminary test-runs of K3, 64 cores on the cloud have been assigned.

Flint K3 is highly compatible with existing web services in order to enhance collaboration among users employing those services. Users can login to Flint K3 with existing web accounts such as Facebook, Twitter, Google, Dropbox, and Evernote. In addition, Flint K3 has its own user management system. Resulting data and image files (graphs) can be sent to Dropbox, Facebook group, Evernote, and Google drive. Of course, those files can be downloaded to the user’s local disk as well.

There are three ways for users to submit simulation jobs to Flint K3. One way is to visit the K3 IFS via a web browser. Users can upload models and configure simulation parameters at that site. The second way is to utilize a link between K3 and various public domain databases. Examples include the PHML model database at Physiome.jp (http://www.physiome.jp/modeldb/) and the SBML model database at BioModels in EMBL-EBI (http://www.ebi.ac.uk/biomodels-main/). Usually each database assigns a unique ID to each model. Users specify such a model ID at K3 IFS, then K3 IFS accesses the database and downloads the model specified by the model ID directly from the database. Once such linkages among databases and simulation servers are built, users can easily check the dynamics of a model in a database. The third way to submit a simulation is to use the REST APIs implemented on IFS, so that applications such as PhysioDesigner and Flint (standalone version) can access K3 directly. For example, a user can submit a simulation job and receive a progress report regarding simulations using the API.



Get Started
(beta release)



Contact: flint __at__ oist.jp

Related projects:
Pasted Graphic 1
Pasted Graphic
© 2013 Open Biology Unit, OIST K3 Administrator