TOP FOLD-RECOGNITION

METHODS EVALUATED SO FAR

(See legend at the bottom of the table).

METHOD GAPS IN RANK % SENS.
ALG SCORING OP EX <9 <5 =1 SCORE
predss+mult-gonnet gloloc raw 9.6 0.60 58 56 52 76% 0.80 NEW
SELEC.: 36/3.7 50/2.6
predss-gonnet gloloc-xg raw 9.6 0.64 57 57 50 74% 0.79 NEW
SELEC.: 26/4.8 46/3.1
predss-gon/4+bow/4+M gloloc raw 1.8 0.15 59 54 50 74% 0.78 NEW
SELEC.: 34/4.3 43/3.2
predss-gonnet gloloc raw 10.8 0.60 58 56 48 71% 0.77
SELEC.: 27/4.7 48/3.1
predss-gon-ln-local local ln 10.8 1.04 59 54 46 68% 0.73
predss-gon/4+bow/4 gloloc raw 2.4 0.15 59 55 45 66% 0.73 NEW
SELEC.: 32/4.7 44/3.2
gon/4+bow/4+Mult gloloc raw 2.4 0.15 55 53 45 66% 0.72 NEW
SELEC.: 32/3.9 39/3.6
elofsson2 gloloc raw 3.2 0.20 53 50 46 68% 0.71
SELEC.: 22/5.3 38/3.9
gonnet/4+bowie/4 gloloc raw 2.4 0.15 57 53 44 65% 0.70 NEW
SELEC.: 35/3.8 45/3.3
predss-blossum gloloc raw 5.2 1.00 55 49 44 65% 0.70
SELEC.: 17/4.4 35/3.0
gonnet Multi gloloc raw 8.4 0.60 55 50 43 63% 0.69 NEW
SELEC.: 29/4.4 40/3.3
gonnet/4+bowie/2 gloloc raw 3.2 0.15 54 50 42 62% 0.69 NEW
SELEC.: 25/4.8 40/3.8
gonnet gloloc-xg raw 8.4 0.64 55 51 42 62% 0.68 NEW
gonnet Multi gloloc raw 10.8 0.60 52 49 41 60% 0.67
SELEC.: 19/5.1 34/3.6
gonnet gloloc raw 10.8 0.60 51 50 40 59% 0.66
SELEC.: 26/4.4 34/3.7
elofsson1 gloloc raw 2.4 0.20 48 43 40 59% 0.63
SELEC.: 15/4.9 38/3.4
gonnet-ln-local local ln 10.8 1.04 47 45 40 59% 0.63
blosum gloloc raw 5.2 1.00 52 45 37 54% 0.61
SELEC.: 17/4.4 29/3.7

LEGEND.
METHOD: Method's name
- predss Uses predicted secondary structure for the probe.
- gonnet Gonnet sequence substitution table.
- blossum Blossum62 sequence substitution table.
- Multi Multiple aligned sequences in the probe.
- other details missing.
ALG: Algorithm used to find the optimum alignment.
- gloloc-xg : global-local, with added gap openning penalty for short diagonals, reference:
- gloloc : global-local, reference:
- global : global (Needleman & Wunsch, no ends penalization), reference:
- local : local (Smith & Waterman), reference:
SCORING: Method used to rank the results.
- raw : sorting by raw scores of the alignments, reference:
- z-scores : sorting by z-scores computed by randomizing the probe sequence, reference:
- ln : sorting by normalized raw scores: raw/ln(length of target), reference:
GAPS: Gap OPEning and EXTension penalties.
IN RANK: Number of correct assignments obtained at rank up to 10, up to 5 and exactly at rank 1.
%: Percentage of correct assignments obtained at rank 1 (out of 68).
SENSITIVITY SCORE: See reference:
SELEC.: For some methods, the number of test sequences before the 1st and 5th false positives are shown. After the "/" the z-score of the 1st and 5th false positive is shown.

OTHER FOLD-RECOGNITION METHODS EVALUATED SO FAR

METHOD			GAPS	IN RANK	%	SENSITIVITY
	ALG	SCORING	OPE EXT	<9 <5 =1		SCORE
predss-identity	gloloc	raw	1.4 0.15	50 45 37	54%	0.61
pam250	gloloc	raw	5.5 1.25	47 44 35	51%	0.59
gonnet-ln-glob	global	ln	13.2 1.04	43 41 36	53%	0.58
gonnet-local	local	raw	10.8 1.04	44 42 37	54%	0.58
gonnet-glob	global	raw	13.2 1.04	39 38 37	54%	0.57
predss-bowie+Mult	gloloc	raw	2.l 0.15	48 43 34	50%	0.57
		SELEC.:	5/6.0	28/3.2
gonnet-ln	gloloc	ln	10.8 1.04	47 43 34	50%	0.57
predss-bowie	gloloc	raw	1.8 0.20	48 42 31	46%	0.56
		SELEC.:	11/4.1	22/3.4
gonnet-ran	gloloc	z-scores	13.2 0.60	46 39 34	50%	0.56
gonnet-ran-local	local	z-scores	13.2 0.16	46 43 34	50%	0.56
gcg	gloloc	raw	4.6 0.20	46 37 31	46%	0.52
bowie +Mult	gloloc	raw	2.l 0.15	45 39 30	34%	0.51
		SELEC.:	7/5.0	20/3.8
gonnet-ran-glob	global	z-scores	13.2 0.16	44 39 29	43%	0.51
identity	gloloc	raw	1.4 0.15	43 40 29	43%	0.50
predss-only	gloloc	raw	0.0 0.70	47 38 26	38%	0.48
bowie	gloloc	raw	1.8 0.20	43 35 25	37%	0.46
		SELEC.:	6/4.5	14/3.6
Variations on bowie	gloloc	raw

For additional evaluations using different methods, alignment algorithms, scoring methods, etc. click here (not yet available).

To include the results of your method in the above table, e-mail me to fischer@mbi.ucla.edu