A homology measure for protein fold classes has been constructed by lo
cally projecting consecutive secondary structures onto a lattice. Taki
ng into account hydrophobic forces we have found a mechanism for forma
tion of domains containing magic numbers of secondary structures and m
ultipla of these domains. We have performed a statistical analysis of
available protein structures and found agreement with the predicted pr
eferred abundances. Furthermore, a connection between sequence informa
tion and fold classes is established in terms of hinge forces between
the structural elements.