Skip to main content

Table 7 Comparison of the present method and seven state ASA predictors on the Manesh dataset

From: Real value prediction of protein solvent accessibility using enhanced PSSM features

Method\Threshold (%) 5 9 10 16 20 25 36 50 60 70 80 90
Rost and Sander --1 74.6 -- 75.0 -- -- -- -- -- -- -- --
Cuff and Barton 79.0 -- -- -- -- 75.0 -- -- -- -- -- --
Manesh et al. -- 75.9 -- 75.5 -- 74.4 74.1 -- -- -- -- --
Ahmad et al. 74.6 -- 71.2 -- -- 70.3 -- 75.9 -- -- -- --
Gianese et al. 75.7 -- 73.4 -- -- 71.6 -- 76.2 -- -- -- --
Nguyen and Rajapakse             
(Two-stage SVM)2 82.9 -- 81.0 -- 78.6 78.1 -- 79.1 83.4 -- -- --
Nguyen and Rajapakse             
(Two-stage SVR)3 81.1 78.7 78.5 77.9 77.6 77.3 76.9 79.5 84.3 89.9 95.0 97.5
Our method 80.9 80.2 80.1 79.4 78.7 78.5 78.4 80.8 85.3 90.7 95.0 97.8
  1. This table reports the accuracy (%) of alternative methods based on the training set of 30 proteins from the Manesh dataset to predict the remaining 185 proteins of Manesh. 1Indicates that the corresponding result was not available from the literature. 2Nguyen and Rajapakse proposed a two-stage SVM approach in 2005 which treats solvent accessibility as a classification problem [19], 3and then proposed a two-stage SVR approach in 2006 which treats solvent accessibility as a regression problem [27].