Since 2004 it has been speculated that the form of Universe might be dodecahedral or even hyperdodecahedral in 4 dimensions.

http://www.youtube.com/watch?v=Nk7SIPljkH0

http://physicsworld.com/cws/article/news/18368

Now there are new theories that the Universe might be Brane (Membrane), in a multidimensional space.

http://www.whillyard.com/science-pages/steinhart-turok.html

Earlier the Hypothesis was that the Universe might be Torus, which ,ight bemost adequate for  Big Bang theory since no centre from where the galaxies and stars are spreading has been ever obsserved.

http://vortexspace.org/display/glossary/Torus+Geometry

Maybe it would be interesting to summarise all observable moving vectors of galaxies  and try to identify if there is any preferible direction. I know that some analysis have been done and that it seems that the (observable) Universe is actualy egg shaped and not spheric at all.

Which theory do you prefer, and are there other hypothesis?

  I have to go with chocolate covered torus! Yummy universe ...

The idea of a Universe that is so far from (spherically) symmetrical (on the large scale) is hard to believe. I recall hearing about it in the Arxiv blog, I think. I don't think this theory is at all mainstream, though. But when I read it first of all, I got the impression it was about a Universe that was not simply connected - there are topologies which on the small scale look normal but which have a quite different large scale structure - for example a Klein Bottle looks like a plane locally. In 3-dimensional topology, there are interesting examples based on dodecahedra where opposite faces are glued together in various ways (you can twist them). If you use the analogy of taking a square and sticking opposite sides together either with or without a twist, this should make sense). So the impression I got several years ago was that someone was claiming the large scale Universe had some exotic topology like that. But perhaps that is a theory for the future.

Ummm...  Big?

They don't know.