Right now, loading of data from SCOP, CATH, and AtomCaches is handled very differently. It would be nice to have 1) an interface that is more consistent between SCOP and CATH, and 2) a more robust way to load arbitrary "structures" given unique identifiers.