Don Berry from MD Anderson has this excellent post on biostatistical/clinical interactions.
http://magazine.amstat.org/blog/2012/02/01/collaborationpolic/
Its worth the read for researchers trying to understand how to operate in a collaborative world.
more...
Not much different in some respects from the model at Duke (of collaborating statisticians working primarily with one discipline). It has a very different funding model; Duke would need to nearly double our faculty to meet that.