Figure 6 shows how the percentage of structural equivalence (i.e., the fraction of the smallest structure that
lies within structurally similar regions) behaves as a function of .
Proteins having detectable sequence similarity (i.e.,
) generally have over
structural
equivalence, or common structural regions. However, as
drops, the percentage of
structural equivalence drops to as low as
. This is a consequence of distantly
related protein structures being similar only within their core secondary structures and having loop/turn
regions that differ substantially. This fraction is somewhat less than that reported by Chothia
&Lesk (1986), who found a minimum of
, though their
protein structure pairs
were more closely related than the
pairs used in this study. For example, the globin fold family
studied by Chothia and Lesk contained only haem containing globins and lacked the more distantly related
phycocyanin and colicin A structures, which are included in this study.