Table 2B.

Equivalence score performance for all methods

						Homologs
	Dali	Matras	Structal	CE	VAST	PSI-BLAST	SGM	SSEARCH
Dali	—	26/12	44/6	59/4	78/4	64/4	76/3	73/4
		0.017	1.6e-8	6.9e-14	0	2.9e-15	0	0
Matras	26/8	—	31/17	52/8	70/7	62/6	72/6	72/6
	0.002		0.03	2.6e-9	1.8e-14	4.1e-13	8.9e-16	8.9e-16
Structal	36/6	28/14	—	46/10	68/11	60/4	69/3	71/5
	1.4e-6	0.022		6.2e/7	1.8e-11	3.7e-14	0	2.2e-16
CE	52/3	46/8	47/21	—	47/25	41/19	58/12	52/13
	7.7e-13	6.9e-8	0.001		0.006	0.003	1.1e-8	5.8e-7
VAST	56/6	49/10	53/28	43/37	—	42/31	59/17	54/20
	1.5e-11	1.4e-7	0.004	0.29		0.12	7.0e-7	4.8e-5
PSI-BLAST	66/2	64/6	73/2	60/14	61/17	—	43/25	39/14
	0	1.2e-13	0	3.1e-8	2.8e-7		0.019	4.0e-4
SGM	73/1	67/5	72/5	59/14	67/14	28/46	—	29/28
	0	3.2e-15	1.1e-16	5.1e-8	9.4e-10	0.024		0.5
SSEARCH	74/3	75/6	76/4	64/10	67/14	41/19	48/21	—
	0	1/1e-16	0	4.5e-11	9.4e-10	0.003	7.8e-4
		Topologs

The upper numbers are the plus (+) and minus (−) values (see Table 2A). The lower numbers are the probability, based upon the binomial distribution, that the two methods perform equivalently. The upper half of the matrix is for the Homolog true positive set, the lower half (shaded), for the Topolog set. The better-performing method is on the left when searching for Homologs, and on the top when searching for topologs (e.g., when searching for Homologs, Dali has a better equivalence number than Structal in 44 families, whereas Structal has a better equivalence number in six families. When searching for Topologs, Dali does better in 36 families, whereas Structal does better in six). The exception is PSI-BLAST vs. SGM, where PSI-BLAST does better with Homologs, but SGM does better with Topologs.