Skip to main content
. 2018 Jun 27;19:82. doi: 10.1186/s13059-018-1450-0

Table 2.

Taxonomic assignment performance (%) and speed of HmmUFOtu and two similar tools, pplacer and EPA, at different taxonomy levels

Dataset Performance measurement Speed
Kingdom Phylum Class Order Family Genus Species
Random TPR 100.00 99.96 99.52 99.22 98.87 95.03 81.96 3.14
TNR nan 96.98 91.76 96.04 98.09 98.09 99.35
PPV 100.00 99.99 99.63 99.43 99.10 95.94 84.85
ACC 100.00 99.94 99.19 98.82 98.62 97.10 98.62
V4 TPR 100.00 99.96 99.52 99.26 99.17 95.78 82.75 4.61
TNR nan 94.85 91.17 95.93 97.98 98.17 99.44
PPV 100.00 99.98 99.60 99.41 99.06 96.09 86.10
ACC 100.00 99.94 99.15 98.84 98.79 97.40 98.77
V1 V3 TPR 100.00 99.97 99.45 99.18 98.93 95.34 82.89 3.22
TNR nan 98.84 89.52 95.51 97.56 97.93 99.26
PPV 100.00 99.99 99.51 99.36 98.86 95.53 82.83
ACC 100.00 99.96 99.01 98.72 98.49 97.11 98.58
V3 V5 TPR 100.00 99.96 99.61 99.39 99.40 96.98 86.71 3.15
TNR nan 98.17 94.86 97.73 99.17 98.97 99.64
PPV 100.00 99.99 99.77 99.68 99.61 97.78 91.05
ACC 100.00 99.96 99.41 99.18 99.33 98.34 99.12
Program Performance measurement Speed
Kingdom Phylum Class Order Family Genus Species
HmmUFOtu TPR 100.00 98.36 95.60 91.63 86.72 83.31 84.02 12.02
TNR nan 95.05 94.97 96.29 97.50 98.71 99.71
PPV 100.00 99.11 95.65 90.81 84.11 80.48 83.04
ACC 100.00 97.86 95.31 94.96 96.07 97.79 99.44
Pplacer TPR 100.00 99.92 99.68 99.58 99.70 100.00 100.00 3.23
TNR nan 69.13 74.95 80.48 87.40 93.56 98.12
PPV 100.00 94.77 82.16 67.11 54.72 49.75 47.74
ACC 100.00 95.25 88.22 85.94 89.03 93.95 98.15
EPA TPR 100.00 99.52 98.58 98.14 97.89 98.00 97.04 0.27
TNR nan 82.12 84.06 87.74 92.06 95.76 98.89
PPV 100.00 96.89 87.74 76.20 65.31 59.53 60.07
ACC 100.00 96.88 91.85 90.71 92.83 95.89 98.86

TPR sensitivity, TNR specificity, PPV precision, ACC accuracy, Nan no observed data

Speed is measured as reads per second per processor. Upper panel: results from running HmmUFOtu on four simulated datasets based on gg_97_otus_GTR database; Lower panel: results from running the three tested programs on the random79 dataset based on gg_79_otus_GTR database. Speed is measured as reads per second per processor. Performance measurement results are based on the aggregate of any replicate samples; speed results are based on the average of any replicate samples