Table 2.
Scores of algorithms when aligning short reads to REF-2
| Algorithm | Species | Genus | ||||
|---|---|---|---|---|---|---|
| F1 | Sensitivity | Precision | F1 | Sensitivity | Precision | |
| Pufferfish | 46.23 | 32.61 | 79.38 | 55.54 | 39.18 | 95.36 |
| Mora + Pufferfish | 47.79 | 33.72 | 82.06 | 56.00 | 39.50 | 96.15 |
| AugPatho2 + Pufferfish | 48.40 | 34.14 | 83.11 | 56.11 | 39.58 | 96.34 |
| Bowtie2 | 64.08 | 52.36 | 82.58 | 74.96 | 61.24 | 96.59 |
| Mora + Bowtie2 | 65.23 | 53.29 | 84.06 | 75.17 | 61.41 | 96.87 |
| AugPatho2 + Bowtie2 | 66.10 | 54.01 | 85.18 | 75.27 | 61.24 | 96.99 |
| Kraken2 | 70.28 | 66.32 | 74.73 | 88.54 | 83.55 | 94.15 |
| Clark* | 13.78 | 7.80 | 59.11 | 20.13 | 11.39 | 86.37 |
On both aligners, Mora performed just a bit worse than AugPatho2, with scores having a difference of at most 1 when using Bowtie2. Mora’s scores are consistently better than the aligners, showing that the re-assignment step is beneficial. Kraken2 was able to map more reads compared to Mora and AugPatho2 due to mapping more reads than Pufferfish and Bowtie2. Clark has an asterisk to signify that its low scores are due to an issue with building the bacteria database