Table 4.

Comparison of receiver operator characteristics (≤10 false positives) for different fold-recognition methods based on all LiveBench8 targets

	Correct predictionsa
	1	2	3	4	5	6	7	8	9	10	Totalb
ORFeusc	83	89	93	93	93	93	93	93	93	93	100
FFAS03c	74	81	83	83	83	83	85	87	88	88	98
FUGUE2.0c	79	79	81	81	82	83	83	83	83	85	90
3D-PSSMc	43	52	62	62	62	63	66	66	72	78	91
DescFoldd	37	56	58	65	70	70	72	73	73	75	86
DescFolde	37	56	58	62	66	66	67	69	69	71	75

^a 1–10: number of correct predictions with higher reliability than the 1–10 false prediction.

^b Total number of correct predictions.

^c The results for ORFeus, FFAS03, FUGUE2.0, and 3D-PSSM are cited from http://bioinfo.pl/Meta/results.pl?B=LiveBench&V=8.

^d The performance was evaluated based on the number of correctly assigned folds.

^e The performance was assessed by the quality of predicted 3D models based on the PSI-BLAST alignment.