Skip to main content

Table 2 Performance of rank-flexible classifiers on 1000 bp query fragments

From: Classifying short genomic fragments from novel lineages using composition and homology

  Ge     Or     Cl     Ph    
  Sn FNr Sp Ur Sn FNr Sp Ur Sn FNr Sp Ur Sn FNr Sp Ur
TACOA 8.3 0.7 91.6 91.0 15.0 1.3 86.1 83.6 18.5 1.2 93.7 80.3 31.2 4.3 87.9 64.4
PhyloPythiaS 35.2 1.6 94.3 63.2 48.8 3.5 92.7 47.7 56.9 4.1 92.8 39.0 58.7 5.6 93.5 35.7
ε-NB (none) a 95.5 2.6 97.9 2.0 96.2 2.5 98.2 1.4 96.7 2.2 98.2 1.0 97.4 1.9 98.8 0.8
ε-NB (strain) b 92.6 4.5 97.5 3.0 92.2 5.1 96.7 2.8 92.3 5.0 96.4 2.7 92.1 5.2 97.5 2.8
ε-NB (species) c 71.0 17.0 84.5 12.0 73.0 16.6 86.2 10.4 75.4 15.7 87.6 8.9 74.8 16.8 89.7 8.4
LCA (strain)b 96.2 0.8 99.3 3.0 95.8 0.9 99.1 3.3 95.4 0.8 99.3 3.9 95.2 0.9 99.3 4.0
LCA (species)c 82.9 4.1 95.5 13.0 85.2 2.4 97.5 12.4 85.6 1.8 98.0 12.6 85.9 1.7 98.7 1.7
LCA+NB (strain) b 91.5 0.1 99.9 8.4 91.0 0.1 99.8 8.9 90.8 0.1 99.9 9.13 90.8 0.1 100 9.1
LCA+NB (species) c 66.9 0.5 99.2 32.6 69.1 0.3 99.5 30.6 70.9 0.2 99.8 28.9 70.4 0.4 99.9 29.2
  1. ε-NB results are for ε = 105; LCA and LCA+NB results are for E-value < 10-5, p = 15%. Ge = genus, Or = order, Cl = class, Ph = phylum
  2. a no models excluded from training set; b strain-level lineages excluded from training set; c species-level lineages excluded from training set.