Table 3.
The F-measures of MetaCluster 5.0, AbundanceBin and BiMeta on samples from S1 to S10
| Samples | MetaCluster 5.0 | AbundanceBin | BiMeta |
|---|---|---|---|
| S1 | 67.11% | - | 98.02% |
| S2 | 88.68% | 72.63% | 60.14% |
| S3 | 71.98% | 83.53% | 97.72% |
| S4 | 77.20% | - | 99.35% |
| S5 | 80.08% | 56.38% | 89.32% |
| S6 | 88.74% | 64.24% | 99.29% |
| S7 | 91.04% | 58.49% | 77.24% |
| S8 | 57.94% | 47.87% | 70.27% |
| S9 | 67.56% | 27.92% | 77.01% |
| S10 | 52.17% | 4.95% | 65.37% |
The symbol “-” indicates that the approaches fail to classify reads on the samples.BiMeta achieves higher F-measure in comparison with that of MetaCluster 5.0 and AbundanceBin for 8 out of 10 samples, while MetaCluster 5.0 gets the highest value for sample S2 and S7 in comparison with that of the remaining approaches.