ModEnzA: Accurate Identification of Metabolic Enzymes Using Function Specific Profile HMMs with Optimised Discrimination Threshold and Modified Emission Probabilities
Table 1
Genome-wide enzyme identification for three bacterial genomes (E. coli, B. aphidicola, and M. pneumoniae) and one eukaryotic genome (P. falciparum) by ModEnzA and EFICAz.
Methods
EFICAz
ModEnzA(Tier I)
ModEnzA(Tier I+II)
Annotation benchmark
HAMAP
E. coli
Sequences
1012
859 (1051)
902 (1021)
930 (1082)
Sensitivity
84.88
89.13
91.89
Specificity
81.73
88.34
85.95
EC numbers
755
653 (728)
663 (697)
699 (775)
Sensitivity
86.49
87.81
92.58
Specificity
89.69
95.12
90.19
B. aphidicola
Sequences
273
257 (273)
264 (271)
265 (273)
Sensitivity
94.13
96.7
97.07
Specificity
94.13
97.41
97.07
EC numbers
245
226 (238)
225 (229)
225 (233)
Sensitivity
92.24
91.83
91.83
Specificity
94.95
98.25
96.56
M. pneumoniae
Sequences
147
119 (149)
126 (139)
126 (139)
Sensitivity
80.95
85.71
85.71
Specificity
79.86
90.64
90.64
EC numbers
127
101 (122)
115 (122)
115 (122)
Sensitivity
79.52
90.55
90.55
Specificity
82.78
94.26
94.26
Annotation benchmark
PlasmoDB
P. falciparum
Sequences
771
341 (480)
350 (415)
358 (431)
Sensitivity
44.22
45.39
46.43
Specificity
71.04
84.33
83.06
EC numbers
410
217 (247)
212 (234)
215 (242)
Sensitivity
52.92
51.7
52.43
Specificity
87.85
90.59
88.84
Numbers within parentheses indicate the total number of sequences or EC numbers identified by each method.
