I was reminded of
http://galaxyproject.org/ when I read this, which is a web front-end that goes between any number of HPC compute resources, data, command-line tools, and scientists.
Users assemble their workflows in a sharable, repeatable and somewhat easy-to-use way using the web-ified versions of their command-line tools. Galaxy takes care of hooking everything up (data, scripts, intermediate results, etc) and provides a UI for watching/controlling the resulting job(s) on the desired HPC cluster.
The goal for this in biology is to address the explosion in data modern biologists have to deal with nowadays - it's often completely overwhelming. Things like Galaxy increase both the accessibility of these command-line tools and also the large HPC compute resources traditionally dominated by users from the physical sciences.
Although it was a few years ago, it was rewarding seeing non-programmer users get onto HPC - importantly, while retaining control of their workflow - after they've struggled for so long on their local hardware. When you see jobs that took weeks or months run in hours or days, it's almost like giving a desert explorer some water.
Systems like Galaxy do require dedicated systems people to support it and make it properly sing though. Gooey could nicely accelerate the web-ification part of adding new software.