Skip to main content

Table 1 Performance of various statistical potentials on models of CASP5-8 experiments.

From: New statistical potential for quality assessment of protein models and a survey of energy functions

Scoring function

models only

native included

 

Average a

Ranked 1 b

Average c

Raw average d

Ranked 1 e

QMEAN6

2.87

85

1.71

3.26

113

QMEAN-all_atom

3.59

74

1.71

2.9

119

QMEAN-SSE_agree

3.74

62

3.72

9.62

39

QMEAN-ACC_agree

4.04

40

3.78

8.83

48

RF_CB_SRS_OD

4.16

61

2.08

3.6

110

RF_CB_OD

4.62

62

2.00

3.65

111

RF_HA_SRS

4.65

49

1.38

1.66

137

RF_CB_SRS

4.72

56

2.18

3.46

114

OPUS_CA

4.72

79

5.13

9.93

55

VSCORE-combined

4.79

53

2.20

3.79

117

QMEAN-pairwise

4.80

54

3.15

5.86

85

Rosetta

5.01

57

4.09

8.03

68

Dong-pair

5.01

58

6.32

14.41

4

RF_CB

5.06

52

2.46

4.31

106

VSCORE-pair

5.08

54

1.85

2.81

128

PROSA-combined

5.11

57

3.38

6.27

87

OPUS_PSP

5.39

54

2.99

4.11

118

RF_HA

5.44

62

2.78

4.37

112

DOPE

5.77

54

3.27

5.97

95

dFIRE

6.03

50

5.69

11.8

33

PROSA-pair

6.03

56

3.54

6.02

95

QMEAN-torsion

6.71

45

3.24

4.66

114

Shortle2006

6.85

35

1.79

2.54

129

Liang_geometric

6.88

44

2.48

3.94

114

QMEAN-solvation

7.32

33

6.27

10.87

54

Shortle2005

7.73

42

3.39

5.19

109

Floudas-CM

7.75

38

7.05

12.77

42

Floudas-Ca

7.79

33

8.36

16.01

10

NAMD 1000

8.06

24

4.96

8.56

78

Melo-ANOLEA

9.62

19

5.19

8.37

86

PC2CA

9.75

19

5.06

8.35

85

Melo-NL

9.99

14

5.85

9.45

80

NAMD 1

11.91

5

10.98

18.04

24

Random¶

9.72

13.9

10.1

10.1

8.3

  1. a The average rank (over 143 decoy sets) in the absence of native structures.
  2. b The number of sets when the best model was ranked as first, in the absence of native structures.
  3. c The average rank when native structures are present.
  4. d The average rank when native structures are present, calculated without compensation for the gap in ranking between experimental structure and first model (see text).
  5. e The number of sets when the best model was ranked as first when native structures are present.
  6. ¶Expected random values were generated by picking a wining model from the decoy sets randomly. Average values over 1000 random trials are shown.
Back to article page

BMC Bioinformatics

ISSN: 1471-2105

Contact us