Skip to main content

Table 3 Improvement of protein secondary structure prediction with the addition of a "residue-specific" RSA threshold for Chou-Fasman and HMM method.

From: Impact of residue accessible surface area on the prediction of protein secondary structures

  Applied Threshold
  Chou-Fasman HMM
  Mean Median Mean Median
A 11.89 10.43 -5.29 -5.41
C 3.33 1.27 4.26 4.93
D 11.73 10.77 5.81 6.66
E 9.16 8.56 -3.55 -3.76
F -0.32 -0.39 1.12 1.55
G 9.11 6.72 12.79 14.25
H 10.92 12.61 2.83 3.28
I -0.01 -1.45 0.08 0.41
K 8.31 5.76 0.25 0.35
L 1.08 1.21 -3.53 -3.49
M 0.17 -0.40 -3.60 -3.62
N 8.38 8.71 7.20 8.12
P 12.32 10.08 11.97 13.56
Q 10.35 9.07 -2.55 -2.53
R 9.67 8.32 -1.21 -1.10
S 11.61 7.89 5.07 5.79
T 1.68 0.16 5.22 6.10
V -0.23 -0.61 2.20 2.50
W -0.57 -0.71 -0.78 -0.84
Y 0.74 0.68 0.87 1.04
Total Improvement 9.99 8.69 3.37 3.92
  Applied Threshold
  Chou-Fasman HMM
  Tertile Mean ± standard deviation Tertile Mean ± standard deviation
A 12.15 12.50 -4.31 -2.69
C 2.64 1.76 3.85 2.93
D 13.61 13.17 8.94 6.01
E 10.35 9.48 -1.97 -1.04
F -0.23 -1.88 -0.17 1.37
G 9.29 8.80 18.48 13.14
H 12.23 11.60 4.48 3.76
I -0.51 0.09 -0.41 0.21
K 8.20 8.57 1.17 1.34
L 0.72 0.49 -3.35 -1.79
M 1.76 -0.81 -1.42 -1.64
N 8.40 8.46 10.72 7.05
P 12.72 14.91 17.33 11.05
Q 10.40 10.65 -0.73 -0.66
R 9.57 9.99 0.28 0.37
S 10.10 13.09 7.41 5.28
T 0.52 0.60 6.44 5.30
V -0.28 -0.20 0.97 2.02
W -0.87 -0.63 -0.50 0.44
Y 0.75 0.92 1.15 1.25
Total Improvement 10.23 10.34 4.32 3.62
  1. The bold-underlined values are those values that show improvements using leave-one-out cross-validation when they are compared with the original method. See the text for more details.