| Method |
A valuable tool in exploiting three-dimensional information
is the databases of protein structure in which domains
with similar three-dimensional architecture are grouped together.
Here, we use the structural classification of proteins
(SCOP)(Conte et al. 2000). In SCOP, protein domains of
known structure that are likely to be homologs are grouped by
an expert into a common superfamily based on their structural
similarity together with functional and evolutionary
considerations. SCOP is widely regarded as an accurate assess-ment of which domains are homologs. However, SCOP remains
subjective and one cannot exclude the possibility that
two domains placed within the same superfamily only share a
common fold as a result of convergent evolution and therefore
are not homologous. |