Skip to main content
. 2012 Feb 8;7(2):e30617. doi: 10.1371/journal.pone.0030617

Table 2. Average of misclassification rates over 100 runs of GFSMLP using the selected propensity (η = 0.2, μ = 0, n = 15, number of iterations = 2,000).

Data setsPropensities AAP872 ABCpred BCPred Combo
1. Hydrophilicity (Parker) 9.43±0.79 5.46±0.77 6.63±0.60 14.02±0.86
2. Accessibility (Emini) 8.87±0.70 4.79±0.48 6.09±0.49 13.24±0.74
3. Flexibility (Karplus) 10.01±0.84 5.34±0.66 6.99±0.74 14.50±1.03
4. Surface Exposed Scale (Janin) 15.80±2.67 6.72±1.05 8.68±1.05 20.31±1.93
5. Polarity (Ponnuswamy) 19.78±1.75 16.59±2.97 13.07±1.84 24.10±2.43
6. Turns (Pellequer) 23.49±2.25 11.24±2.08 16.25±2.64 26.37±2.06
7. Antigenicity (Kolaskar) 11.00±1.28 5.84±0.62 8.06±0.96 16.96±1.29
8. Beta-turn (Chou) 8.40±0.60 4.53±0.45 6.05±0.52 12.67±0.66