Real value prediction of protein solvent accessibility using enhanced PSSM features - BMC Bioinformatics
- ️Wu, Chih-Peng
- ️Fri Dec 12 2008
Table 1 The recent developments, in chronological order, for real value ASA prediction
From: Real value prediction of protein solvent accessibility using enhanced PSSM features
Work | Regression tool | Description of features | MAE (%)1 |
|---|---|---|---|
Ahmad et al., 2003 | NN2 | Amino acid composition | 18.8 |
Yuan and Huang, 2004 | SVR3 | Amino acid composition | 18.5 |
Adamczak et al., 2004 | NN | PSSM4 | 15.3–15.85 |
Wang et al., 2005 | MLR6 | Amino acid composition, PSSM and sequence length | 16.2 |
Garg et al., 2005 | NN | PSSM and secondary structure information | 15.9 |
Nguyen and Rajapakse, 2006 | Two-stage SVR | PSSM | 15.7 |
- 1Mean absolute error of real RSA values. All the methods were evaluated with a three-fold cross-validation on the Barton dataset, except Adamczak et al. used their own dataset. 2Neural network. 3Support vector regression. 4Position specific scoring matrix. 5The MAEs reported in this work were evaluated on a different dataset to other works. 6Multiple linear regression.