Alternatively, the abstract space of
protein folds. That is, given all possible sequences of
amino acids (theoretically infinite), the space of
structures that they can fold into.
Of course, there is a practical upper limit on the length of sequence - most proteins are 100-1000 residues. Also, there are only so many sequences (out of the many, many possible ones) that have been 'attempted' - or synthesised. However, it is an interesting thought : Is fold space infinite? (that is, as sequences grow beyond physical limits, could they be folded in silico into new and different structures?)
Luckily, fold space is degenerate - several sequences can correspond to a single 'fold'. The concept of a fold is somewhat fuzzy, of course, but the definition could be made definate using RMSDs. This is what allows organisms to search through the space of folds using an evolutionary approach. If point mutations caused large changes in fold, it would make it more difficult to evolve folds.