Table 3.
Ec
–
At
|
Ec
–
Sc
|
Sc
–
At
|
|||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1e-20 | 1e-16 | 1e-12 | 1e-8 | 1e-4 | 1e-20 | 1e-16 | 1e-12 | 1e-8 | 1e-4 | 1e-20 | 1e-16 | 1e-12 | 1e-8 | 1e-4 | |
Precision [%] |
73.3 |
85.6 |
86.7 |
90.0 |
90.8 |
78.6 |
78.6 |
81.6 |
86.7 |
92.9 |
84.7 |
85.4 |
85.4 |
97.5 |
93.7 |
Recall [%] |
73.3 |
85.6 |
86.7 |
90.0 |
92.1 |
78.6 |
78.6 |
81.6 |
86.7 |
92.9 |
84.7 |
85.4 |
85.4 |
97.5 |
96.0 |
F-measure[%] |
73.3 |
85.6 |
86.7 |
90.0 |
91.4 |
78.6 |
78.6 |
81.6 |
86.7 |
92.9 |
84.7 |
85.4 |
85.4 |
97.5 |
94.7 |
Unclassified [%] | 76.7 | 52.2 | 40.0 | 33.3 | 17.8 | 21.4 | 21.4 | 18.4 | 13.3 | 7.1 | 15.3 | 14.6 | 14.6 | 2.5 | 2.5 |
HMMER prediction results (sequence E-values) under the given E-value confidence thresholds. The results were normalized by the size of the reference database (see text). HMMER gave a better accuracy under loose thresholds compared to BLAST.