Skip to main content

Table 1 Performance comparison in native recognition

From: ANDIS: an atomic angle- and distance-dependent statistical potential for protein structure quality assessment

Decoy sets

CASP5–8

CASP10–13

I-TASSER

3DRobot

Rosetta

No. totald

No. of targetsa

143 (2759)

175 (13,474)

56 (24,707)

200 (60,200)

58 (5858)

632 (106,998)

Dfireb

64 (0.61)

56 (0.72)

43 (2.80)

1 (0.83)

22 (1.55)

186 (0.99)

RW

65 (1.01)

36 (0.86)

53 (4.42)

0 (−0.30)

20 (1.48)

174 (0.90)

GOAP

106 (1.67)

89 (1.62)

45 (4.98)

94 (1.85)

45 (3.38)

379 (2.16)

DOOP

135 (1.96)

121 (1.99)

52 (6.18)

197 (3.53)

50 (3.91)

555 (3.02)

ITDA

71 (1.15)

117 (1.67)

52 (4.98)

196 (3.83)

53 (3.52)

489 (2.70)

VoroMQA

132 (2.00)

111 (1.77)

48 (5.11)

114 (1.89)

43 (3.09)

448 (2.28)

SBROD

88 (1.62)

119 (2.32)

33 (3.25)

49 (1.76)

42 (3.02)

331 (2.13)

AngularQA

59 (1.26)

24 (1.11)

29 (1.82)

9 (0.99)

2 (0.12)

123 (1.08)

ANDISc

138 (2.16)

129 (2.32)

47 (6.45)

200 (4.99)

50 (4.27)

564 (3.67)

  1. aThe total number of structures (including native structures) are given in parentheses
  2. bThe number of proteins whose native structure is given the lowest energy score by the potential are listed outside the parentheses. The average Z-scores of native structures are listed in parentheses. Z-score is defined as (<Edecoy> − Enative)/δ, where Enative is the energy score of native structure, <Edecoy> and δ are respectively the average and the standard deviation of energy scores for all decoys in the set. But Z-score for VoroMQA energy score is calculated by (Enative −  < Edecoy>)/δ, so that Z-scores of native structures for all potentials are “the higher the better”
  3. cCalculation is based on a distance cutoff of 7.0 Å
  4. dZ-scores are calculated by averaging over all 632 decoy sets
Back to article page

BMC Bioinformatics

ISSN: 1471-2105

Contact us