Table 1 Quantitative performance comparisons for myristoylation prediction algorithms. Each model was evaluated using the same test set, consisting of 80 positive and 185 negative plant sequences. For the Expasy algorithm, only 183 negative sequences could be analyzed because two contained "X" characters (denoting ambiguous amino acids), which the program was unable to process. Abbreviations: TP, true positives; FN, false negatives; FP, false positives; TN, true negatives. Names in parentheses after each HMM indicate the sequence weighting scheme used to build the model.
From: Predicting N-terminal myristoylation sites in plant proteins
Model Name | Threshold Cutoff | Number Correct | TP | FN | FP | TN | Accuracy (TP+TN)/TOTAL | Coverage TP/(TP+FN) |
|---|---|---|---|---|---|---|---|---|
Prosite | (nominal) | 238 | 64 | 16 | 11 | 174 | 89.8% | 80.0% |
NMT | -2.75 | 252 | 70 | 10 | 3 | 182 | 95.1% | 87.5% |
NMT | 0.00 | 233 | 51 | 29 | 3 | 182 | 87.9% | 63.8% |
BGM | 1.85 | 248 | 77 | 3 | 14 | 171 | 93.6% | 96.3% |
BGM | 0.00 | 236 | 79 | 1 | 28 | 157 | 89.1% | 98.8% |
Expasy | 0.89 | 244 | 62 | 18 | 1 | 182 | 92.8% | 77.5% |
Expasy | 0.85 | 241 | 62 | 18 | 4 | 179 | 91.6% | 77.5% |
Expasy | 0.40 | 220 | 66 | 14 | 29 | 154 | 83.7% | 82.5% |
HMM80 (Gerstein) | 2.05 | 265 | 80 | 0 | 0 | 185 | 100.0% | 100.0% |
HMM80H (Henikoff) | 1.55 | 265 | 80 | 0 | 0 | 185 | 100.0% | 100.0% |
HMM80V (Voronoi) | 1.75 | 265 | 80 | 0 | 0 | 185 | 100.0% | 100.0% |