I am in the same situation as you are. The basic documentation is at
http://wiki.g2.bx.psu.edu/Admin/Config/Performance/Cluster#Submitting_Jobs_as_the_Real_User. However, there could be some site-specific issues with that configuration. For instance the user names in my instance do not correspond to the uid part of the galaxy username (email address), which is what released galaxy-dist code assumes. Fortunately, Nate and Scott from the Core Galaxy Team helped me get through almost all of the issues while we were at the Galaxy Community Conference. I am still working through the single remaining issue of drmaa not passing modified resource requests to Torque, but the rest appears to work stably with Torque 2.5.12.