The GenApp framework integrated with Airavata for managed compute resource submissions

A new framework (GenApp) for rapid generation of scientific applications running on a variety of systems including science gateways has recently been developed. This framework currently builds a GUI and/or web-based user interface for a variety of target environments on a collection of executable modules. The method for execution of modules has limited framework restrictions: primarily the requirement of wrapping the application to accept input and output formatted in JavaScript Object Notation (JSON). Initial implementation supports direct execution on a user's workstation, a web server, or a compute resource accessible from the web server. After a successful initial workshop utilizing the framework to create a web-based user interface wrapping a scientific software suite, it was discovered that long-running jobs would sometimes fail, because of the loss of a Transmission Control Protocol (TCP) connection. This precipitated an improvement to the execution method with the bonus of easily allowing multiple web clients to attach to the running job. To support a diversity of queue managed compute resources, a Google Summer of Code' project was completed to integrate the Apache Airavata middleware as an additional execution model within the GenApp framework. New features of file management, job management with progress, and message box support are described. Concurrency and Computation: Practice and Experience, 2015.(c) 2015 Wiley Periodicals, Inc.