Given that proteins are fundamentally asymmetric (their chain has a unique directionality, and most of the chain links have a handedness), it may seem surprising that a majority of natural proteins forms highly symmetrical complexes involving two or more protein subunits.
In a "dispatch" in the current issue of Current Biology, Kevin Plaxco and I have commented on recent work that manages to explain this natural tendency towards symmetry at least for dimers:
Protein Complexes: The Evolution of Symmetry
Current Biology, Volume 19, Issue 1, R25-R26, 13 January 2009
(summary is free, access to pdf requires subscription or institutional access)