Fold assignments for proteins from the Helicobacter pylori genome are carri
ed out using BASIC, a profile-profile alignment algorithm recently tested o
n the Mycoplasma genitalium and Escherichia coli genomes, The fold assignme
nts are followed by automated function evaluation, based on the multilevel
description of functional sites in proteins. Over 40% of the proteins encod
ed in the H, pylori genome can be recognized as belonging to a protein fami
ly with known structure. Previous estimates suggested that only 10-15% of g
enome proteins could be characterized this way This dramatic increase in th
e number of recognized homologies between H. pylori proteins and structural
ly characterized protein families is partly due to the rapid increase of th
e database of known protein structures, but mostly it is due to the signifi
cant improvement in prediction algorithms. Knowledge of a protein fold adds
a nem dimension to our understanding of its function and, similarly struct
ure prediction can also add to understanding, verification, and/or predicti
on of function for uncharacterized proteins. Several examples analyzed in m
ore detail in this article illustrate insights that can be achieved from st
ructure and detailed function prediction. Proteins 1999;36:20-30, (C) 1999W
iley-Liss, Inc.