TABLE III.

Classification Accuracy and Measures of Linear Correlation Between Structural and Sequence Similarity for Homologous and Analogous Pairs for 16 Superfolds

SCOP Fold Name	Discrimination Accuracy	ρ(Rmsd) for Homologs	ρ(LHM) for Homologs	ρ(Rmsd) for Analogs	ρ(LHM) for Analogs
Cupredoxin-like	98/58	−0.59*	−0.72*	0.63	−0.81*
β-Trefoil	64/95	−0.90*	−0.74*	−0.15	−0.42*
OB-fold	56/79	−0.52*	−0.52*	−0.26*	−0.35*
TIM β/α-barrel	22/99	−0.70*	−0.63*	−0.23*	−0.27*
Profilin-like	94/97*	−0.82*	−0.79*	−0.32	−0.25
β-Grasp (ubiquitin-like)	95/90*	−0.70*	−0.49*	−0.21*	−0.19*
Adenine nucleotide α hydrolase-like	71/75*	−0.49*	−0.42*	0.06	0.19
Thioredoxin fold	100/5	−0.61*	−0.62*	0.007	−0.18
Ribonuclease H-like motif	35/89	−0.71*	−0.73*	−0.11	0.14
Ntn hydrolase-like	100/92*	−0.81*	−0.73*	0.06	−0.1
Flavodoxin-like	83/97*	−0.66*	−0.59*	−0.18*	−0.08
Four-helical up-and-down bundle	84/100*	−0.67*	−0.48*	0.002	−0.06
Immunoglobulin-like β-sandwich	63/96	−0.54*	−0.52*	−0.18*	−0.05*
SH3-like barrel	98/98*	−0.35*	−0.40*	−0.21*	0.04
Cystatin-like	92/100*	−0.34	−0.36	0.25	−0.02
Ferredoxin-like	82/100*	−0.46*	−0.38*	−0.07	−0.02

The first column reports percentage of correctly assigned homologous–analogous pairs using a linear combination of 6 main discriminant parameters. The second and third columns show the Pearson correlation coefficients for the dependencies of RMSD and LHM on sequence identity calculated for homologous pairs. The last two columns list Pearson correlation coefficients for the same type of dependencies calculated for analogous pairs of each superfold. Asterisks next to the classification parameters indicate those cases where more than two thirds of homologs and analogs are classified correctly. The folds that have statistically significant linear correlation defined by P-value ≤ 0.01 are indicated by the asterisks next to their correlation coefficients.