Table 1.
Virus recovery of VIBRANT, VirFinder, and VirSorter from mixed metagenomes and a virome
| Metagenome | Seqs. total (≥ 1 kb) | Seqs. ≥ 4 ORFs | Metric | VIBRANT | VirFinder (score ≥ 0.90) | VIBRANT vs. VirFinder | VirSorter(cat. 1 and 2) | VIBRANT vs. VirSorter |
|---|---|---|---|---|---|---|---|---|
| Human gut: adenoma | 34,883 | 11,360 | Total putative viruses | 527 | 604 | 0.87 | 284 | 1.86 |
| Total virus length (bp) | c5,234,242 | 1,696,118 | 3.09 | 3,982,292 | 1.31 | |||
| Total virus proteins | 7,661 | 2,134 | 3.59 | 5,484 | 1.40 | |||
| Human gut: carcinoma | 53,946 | 18,669 | Total putative viruses | 784 | 1,329 | 0.59 | 450 | 1.74 |
| Total virus length (bp) | 5,611,953 | 3,500,838 | 1.60 | 4,182,862 | 1.34 | |||
| Total virus proteins | 8,401 | 4,644 | 1.81 | 5,945 | 1.41 | |||
| Human gut: healthy | 42,739 | 17,079 | Total putative viruses | 565 | 672 | 0.84 | 309 | 1.83 |
| Total virus length (bp) | 5,623,082 | 2,411,049 | 2.33 | 4,512,571 | 1.25 | |||
| Total virus proteins | 8,202 | 3,230 | 2.54 | 6,127 | 1.34 | |||
| Thermophilic compost | 68,815 | 21,620 | Total putative viruses | 1,047 | 878 | 1.19 | 383 | 2.73 |
| Total virus length (bp) | 10,253,162 | 2,238,129 | 4.58 | 3,290,654 | 3.12 | |||
| Total virus proteins | 9,912 | 2,806 | 3.53 | 4,400 | 2.25 | |||
| Freshwater lake (bog) | 79,862 | 26,832 | Total putative viruses | 5,626 | 7,567 | 0.74 | 1,503 | 3.74 |
| Total virus length (bp) | 34,976,570 | 25,357,664 | 1.38 | 15,436,797 | 2.27 | |||
| Total virus proteins | 56,120 | 37,537 | 1.50 | 21,280 | 2.64 | |||
| *Estuary virome | 5,247 | 3,277 | Total putative viruses | 3,141 | 2,294 | 1.37 | 1,121 | 2.80 |
| Total virus length (bp) | 6,591,285 | 6,478,804 | 1.02 | 5,163,674 | 1.28 | |||
| Total virus proteins | 20,500 | 12,035 | 1.70 | 9,645 | 2.13 |
Mixed community assembled metagenomes from human gut, thermophilic compost, and freshwater, as well as an estuary virome, were used to compare virus prediction ability between the three programs. For each assembly, the scaffolds were limited to a minimum length of 1000 bp. Only a subset of each dataset contained scaffolds encoding at least four open reading frames. VIBRANT, VirFinder (score minimum of 0.90), and VirSorter (categories 1 and 2) were compared by total viral predictions, total combined length of predicted viruses, and total combined proteins of predicted viruses. Comparison columns, denoted “VIBRANT vs. VirFinder” and “VIBRANT vs. VirSorter”, display the comparison ratio of the given metric; bold indicates greater performance by VIBRANT. The asterisk represents that VIBRANT, VirFinder and VirSorter were ran with alternate settings (see Methods)