Table 1. Percentage of bases correctly assigned to modeled taxa by different methods for the AMD metagenome scaffolds.
Rank | PhyloPythiaS sample-specific | PhyloPythiaS generic | BLASTN | MEGAN | NBC |
Genus | 41.353 | 0.000 | 0.000 | 0.000 | 0.000 |
Family | 41.353 | 0.000 | 1.685 | 0.000 | 0.000 |
Order | 74.706 | 38.189 | 45.536 | 42.210 | 1.742 |
Class | 74.706 | 38.189 | 45.536 | 42.210 | 1.742 |
Phylum | 89.540 | 47.821 | 47.011 | 42.673 | 1.798 |
Domain | 92.673 | 88.978 | 86.042 | 70.194 | 44.805 |
The reference taxonomic affiliations were obtained by aligning the test scaffolds with the draft genomes. For PhyloPythiaS (both generic and sample-specific), the drop in accuracy is mostly due to unassigned sequences at a particular rank, while other methods produced more false assignments. Thermoplasmatales archaeon Gpl (comprising 21.8% of the total bases) has no defined parental clade at the genus and family ranks, contributing to the observed lower accuracy values for these ranks. Additional measures are shown in Figure S6.