Table 3. Weights of tri-peptides in the optimal SVM model.
Tri-Peptide | Rank | Weight Score* | Tri-Peptide | Rank | Weight Score* |
QQP | 1 | 503251.79 | GQQ | 11 | 121677.62 |
PQQ | 2 | 488627.71 | QPY | 12 | 116598.60 |
QPQ | 3 | 367386.40 | YPQ | 13 | 113237.37 |
QPF | 4 | 246462.39 | QQF | 14 | 81709.59 |
FPQ | 5 | 234868.65 | PYP | 15 | 79191.37 |
PQP | 6 | 231353.73 | FQQ | 16 | 77357.97 |
QGQ | 7 | 153161.76 | PPP | 17 | 76320.05 |
PFP | 8 | 151840.02 | QPP | 18 | 64756.05 |
QQQ | 9 | 128930.20 | QFP | 19 | 63814.16 |
QQG | 10 | 122291.90 | PPQ | 20 | 63173.33 |
Weight scores are calculated by the formula w = ∑ α i xi. Here α is dual representation of the decision boundary; and xi (i = 0, 1, 2…n) is vector described in SVM model. Both α i and xi are available in model file.