Skip to main content

Table 8 CASP7 results, [6, 11] residue separation

From: Ab initio and template-based prediction of multi-class distance maps by two-dimensional recursive neural networks

template ID

[0,10)%

[10,20)%

[20,30)%

[30,40)%

[40,50)%

[50,60)%

[60,70)%

[70,80)%

All

proteins

29

29

16

4

7

4

3

1

93

positive pairs

3318

1990

1696

772

636

744

254

158

9568

negative pairs

60204

49856

43628

13682

13416

13146

3838

3112

200882

SAM_T06

12.2%

8.4%

9.4%

18.6%

12.5%

12.4%

11.5%

9.4%

11.0%

Betapro

26.0%

20.2%

27.8%

48.2%

26.9%

23.2%

27.9%

19.4%

25.7%

ProfCon

34.1%

31.4%

29.7%

40.2%

32.7%

32.3%

37.7%

41.2%

33.3%

Possum

17.2%

15.0%

18.5%

22.2%

21.5%

13.6%

18.3%

24.0%

17.4%

SVMcon

25.6%

25.0%

20.1%

23.6%

22.7%

25.9%

32.0%

36.7%

24.7%

8 AI

37.0%

30.2%

31.7%

47.6%

45.8%

48.7%

39.6%

59.6%

37.2%

8 TE

34.5%

41.6%

54.2%

78.4%

77.7%

91.1%

95.8%

92.8%

53.3%

  1. Results for the CASP7 targets: sequence separations of 6 to 11 residues, inclusive. Comparison between our two predictors (8 AI and 8 TE ) and the predictors ranked highest at CASP7. We report F1 for the contact class, as a function of sequence similarity to the best PSI-BLAST template. Predictions are from the CASP7 web site. ProfCon is in italics because it predicted 73 out of 93 maps, hence its results are not exactly comparable. In bold is the highest F1 of all predictors, excluding ProfCon, and excluding 8 TE , which uses templates and has the highest F1 in all ranges except [0,10)% where it is slightly worse than only 8 AI .