Table 4. Comparison of the number of rounds and final representatives when modifying the distance metric and/or the dividing scheme for parallel processing.
Five replicates of each combination were carried out for the random sort, whereas the taxonomic sort is deterministic. JI-based (direct) analyses were run using a distance threshold of 0.84, where IGF-based (direct) analyses used a threshold of 0.66. Pack size was 200 and the clustering mode was set to “loose”.
| Dataset | dist./appr. | sort | # rounds | # repr. |
|---|---|---|---|---|
| Bacteria | JI-d | taxonomic | 4 | 836 |
| Bacteria | JI-d | random | 18 | 902 |
| Bacteria | JI-d | random | 17 | 903 |
| Bacteria | JI-d | random | 17 | 894 |
| Bacteria | JI-d | random | 18 | 915 |
| Bacteria | JI-d | random | 17 | 908 |
| Bacteria | IGF-d | taxonomic | 4 | 702 |
| Bacteria | IGF-d | random | 10 | 435 |
| Bacteria | IGF-d | random | 10 | 458 |
| Bacteria | IGF-d | random | 10 | 456 |
| Bacteria | IGF-d | random | 9 | 438 |
| Bacteria | IGF-d | random | 10 | 493 |
| Proteobacteria | IGF-d | taxonomic | 3 | 165 |
| Proteobacteria | IGF-d | random | 3 | 115 |
| Proteobacteria | IGF-d | random | 3 | 105 |
| Proteobacteria | IGF-d | random | 3 | 100 |
| Proteobacteria | IGF-d | random | 3 | 124 |
| Proteobacteria | IGF-d | random | 3 | 114 |
| Firmicutes | IGF-d | taxonomic | 4 | 333 |
| Firmicutes | IGF-d | random | 4 | 190 |
| Firmicutes | IGF-d | random | 5 | 212 |
| Firmicutes | IGF-d | random | 5 | 224 |
| Firmicutes | IGF-d | random | 4 | 194 |
| Firmicutes | IGF-d | random | 4 | 172 |