Table 3 Translation initiation site prediction correctness (TIS correctness) and complete/incomplete classifi-cation accuracy (Gene Type Accuracy) of the Neural Net and MetaGene according to GenBank annotation. Performance was measured on 700 bp fragments that were randomly excised from each test genome to 5-fold coverage (mean and standard deviation for 10 replicates per species are given).
From: Gene prediction in metagenomic fragments: A large scale machine learning approach
| Â | TIS CORRECTNESS | GENE TYPE ACCURACY | ||
|---|---|---|---|---|
Species | Neural Net | MetaGene | Neural Net | MetaGene |
Archaeoglobus fulgidus | 69.8 ± 0.32 | 73.6 ± 0.32 | 98.1 ± 0.05 | 97.2 ± 0.07 |
Methanococcus jannaschii | 69.4 ± 0.52 | 73.3 ± 0.52 | 99.0 ± 0.09 | 97.6 ± 0.12 |
Natronomonas pharaonis | 75.2 ± 0.58 | 82.9 ± 0.28 | 96.9 ± 0.16 | 97.6 ± 0.09 |
Buchnera aphidicola | 86.5 ± 0.40 | 88.6 ± 0.64 | 99.1 ± 0.09 | 98.3 ± 0.21 |
Burkholderia pseudomallei | 70.1 ± 0.45 | 73.0 ± 0.28 | 97.6 ± 0.08 | 96.9 ± 0.09 |
Bacillus subtilis | 79.7 ± 0.32 | 66.1 ± 0.42 | 98.6 ± 0.05 | 97.0 ± 0.08 |
Corynebacterium jeikeium | 78.2 ± 0.49 | 73.4 ± 0.68 | 98.1 ± 0.08 | 96.6 ± 0.11 |
Chlorobium tepidum | 68.1 ± 0.46 | 71.9 ± 0.45 | 98.1 ± 0.08 | 96.7 ± 0.13 |
Echerichia coli | 84.5 ± 0.31 | 78.2 ± 0.15 | 98.7 ± 0.06 | 97.0 ± 0.08 |
Helicobacter pylori | 87.3 ± 0.40 | 77.1 ± 0.33 | 99.2 ± 0.09 | 96.4 ± 0.16 |
Pseudomonas aeruginosa | 78.4 ± 0.22 | 81.0 ± 0.36 | 97.7 ± 0.03 | 97.2 ± 0.07 |
Prochlorococcus marinus | 86.6 ± 0.40 | 88.6 ± 0.47 | 99.0 ± 0.07 | 97.8 ± 0.10 |
Wolbachia endosymbiont | 79.3 ± 0.77 | 79.9 ± 0.42 | 98.7 ± 0.13 | 96.9 ± 0.17 |