Skip to main content

Table 3 Performance comparisons between the potentials we built and several widely-used statistical potentials

From: Diverse effects of distance cutoff and residue interval on the performance of distance-dependent atom-pair potential in protein structure prediction

Decoy sets

Measurements

Dfire

RW

GOAP

ave-6-6d

rw-17-3e

Bestf

I-TASSER

R1-numa

43

53

45

42

41

56 (6)

Z-scoreb

2.80

4.42

4.98

2.42

2.97

11.21 (dope-5-0)

PCCc

−0.47

−0.50

−0.50

−0.09

−0.51

−0.55 (rw-15/16–0)

Moulder

R1-num

18

19

19

19

19

20 (89)

Z-score

2.67

2.78

3.48

2.97

2.75

8.17 (rw-8-0)

PCC

−0.84

−0.83

−0.88

−0.52

−0.88

−0.89 (rw-16-2)

Rosetta

R1-num

22

20

45

41

18

48 (ave-8-14, srs-6-8)

Z-score

1.55

1.48

3.38

3.11

1.46

3.56 (srs-6-7)

PCC

−0.37

−0.36

−0.51

−0.31

−0.36

−0.45 (srs-6-13/15)

3DRobot

R1-num

1

0

94

176

19

184 (ave-5-5/6)

Z-score

0.83

−0.30

1.85

3.19

1.16

3.50 (ave-5-5)

PCC

−0.86

−0.85

−0.90

−0.70

−0.86

−0.88 (ave-19/20/21–5)

CASP10

R1-num

26

16

41

53

31

55 (ave-7-6/7/8)

Z-score

0.76

0.86

1.60

1.34

1.31

1.70 (dope-6-10/11/12)

PCC

−0.40

−0.41

−0.53

−0.22

−0.54

−0.56 (rw-18-3, rw-19-4)

CASP11

R1-num

24

15

37

47

33

49 (14)

Z-score

0.82

1.01

1.91

1.37

1.50

1.72 (dope-6-11)

PCC

−0.36

−0.40

−0.54

−0.23

−0.52

−0.52 (rw-17-3)

Total/Average

R1-num

134

123

281

378

161

378 (ave-6-6)

Z-score

1.20

0.95

2.40

2.55

1.55

2.66 (dope-5-5)

PCC

−0.60

−0.61

−0.68

−0.43

−0.66

−0.66 (rw-17/18–3)

  1. aThe number of decoy sets whose native structure is given the lowest energy score by the potential
  2. bDefined as (<E decoy > − E native)/δ, where E nativeis the energy score of native structure, <E decoy > and δare respectively the average and the standard deviation of energy scores of structural decoys
  3. cThe average Pearson’s correlation coefficient between the energy score and TM-score of all structures in each decoy set, including the native structure
  4. dThe potential based on the averaging reference state with both distance cutoff and residue interval to be 6
  5. eThe potential based on the random-walk chain reference state with distance cutoff = 17 and residue interval = 3
  6. fThe best values among the results of all 1728 potentials with different reference states, distance cutoffs and residue intervals. The corresponding potentials that achieve this values are given in parentheses (e.g. rw-15/16–0 means the potentials rw-15–0 and rw-16-0). Only the number of potentials is given in parentheses if more than 3 potentials can achieve the best value