Analysis of CASP2 protein threading results shows that the success rat
e of structure predictions varies widely among prediction targets, We
set ''critical'' thresholds in fold recognition specificity and thread
ing model accuracy at the points where ''incorrect'' CASP2 predictions
just outnumber ''correct'' predictions. Using these thresholds we fin
d that correct predictions were made for all of those targets and for
only those targets where more than 50% of target residues may be super
imposed on previously known structures, Three-fourths of these correct
predictions were furthermore made for targets with greater than 12% r
esidue identity in structural alignment, where characteristic sequence
motifs are also present. Based on these observations we suggest that
the sustained performance of threading methods is best characterized b
y counting the numbers of correct predictions for targets of increasin
g ''difficulty'' We suggest that target difficulty may be assigned, on
ce the true structure of the target is known, according to the fractio
n of residues superimposable onto previously known structures and the
fraction of identical residues in those structural alignments. (C) 199
8 Wiley-Liss, Inc.