Table 6.
Accuracy values obtained with RF-classifiers tested on different datasets
| Genome | Condition | Original dataset | Different condition | Different genomes/conditions | |
|---|---|---|---|---|---|
| |
|
EC1 |
EC2 |
SE1 |
S2 |
|
E. coli |
EC1 |
0.98 |
0.84 (0.001) |
0.83 (0.004) |
0.83 (0.01) |
| |
|
EC2 |
EC2 |
SE1 |
SE2 |
| |
EC2 |
0.98 |
0.86 (0.001) |
0.83 (0.004) |
0.81 (0.01) |
| |
|
SE1 |
SE1 |
EC1 |
EC2 |
|
S. enterica |
SE1 |
0.97 |
0.93 (0.001) |
0.92 (0.01) |
0.91 (0.001) |
| |
|
SE2 |
SE2 |
EC1 |
EC2 |
| SE2 | 0.98 | 0.87 (0.001) | 0.85 (0.02) | 0.86 (0.004) | |
The accuracy values obtained performing a classifier trained on one dataset and tested on different data sets of different conditions and different organisms. Each test was conducted running the RF-based classifier on 50 bootstrap samples compiled from a different dataset and using the corresponding accuracy values to execute a t-test. All the p-values were less than 0.001.