Table 1. Results of supervised classification analysis on the different sets of proteins before and after filtering.
| Classification accuracy (%) before filtering | Classification accuracy (%) after filtering |
||||
|---|---|---|---|---|---|
| Fisher’s ANOVA | Runs | ReliefF | 29-protein set | ||
| No. of features | 499 | 81 | 6 | 37 | 29 |
| Classification algorithm | |||||
| k-nearest neighbors | 63.2% | 68.4% | 73.7% | 68.4% | 84.2% |
| Multilayer perceptron neural network | 52.6% | 94.7% | 68.4% | 79.0% | 94.7% |
| Radial basis function neural network | 79.0% | 68.4% | 63.2% | 79.0% | 84.2% |
| Support vector machine | 79.0% | 73.7% | 84.2% | 100% | 100% |
Classification accuracy was estimated as the overall number of correctly classified samples divided by the total number of samples through a leave-one-out cross-validation procedure. The highest classification accuracy achieved by each of the five classification algorithms is shown in bold.