Table 1.
Taxonomic assignment of contigs (human fecal data, progressive assembly) and classification precision and sensitivity.
| Subfamily | Profile HMM seed | ||||
|---|---|---|---|---|---|
| VP1R1 | VP1R4 | VP4R1 | VP4R3 | Totala | |
| Alpavirinae | 47 | 43 | 38 | 34 | 65 |
| Gokushovirinae | 11 | 0 | 11 | 10 | 17 |
| Pichovirinae | 1 | 0 | 1 | 1 | 1 |
| Gokush/Alpab | 0 | 0 | 1 | 2 | 2 |
| Total | 59 | 43 | 51 | 47 | 85 |
| VP1R4-positive | 40 | 43 | 23 | 24 | 49 |
| Sensitivity for Alpavirinae | (47/65) 72.31% | (43/65) 66.15% | (38/65) 58.46% | (34/65) 52.31% | |
| Precision for Alpavirinae | (47/59) 79.66% | (43/43) 100.00% | (38/51) 74.51% | (34/47) 72.34% | |
| Sensitivity for VP1R4 | (40/49) 81.63% | (43/49) 87.76% | (23/49) 46.94% | (24/49) 48.98% | |
| Precision for VP1R4 | (40/59) 67.80% | (43/43) 100.00% | (23/51) 45.10% | (24/47) 51.06% | |
Contigs generated by GenSeed-HMM with the respective profile HMM (VP1R1, VP1R4, VP4R1, and VP4R3) were compared against all reference Microviridae proteins (Roux et al., 2012) using blastx with a cutoff E-value of 1e-20. When hits with similar scores were obtained to more than one subfamily, taxonomic assignment was set to two subfamilies. Contigs were also evaluated for the presence of VP1R4 region by hmmsearch, and the number of positive contigs is shown.
Total number of De-replicated contigs (See Figure 3) that belonged to a given taxonomic assignment.
Represents a set of two contigs where the best BLAST hit annotation was below the E-value cutoff and they were equally distant by percent identity to Gokushovirinae and Alpavirinae, so no single assignment was possible.