Table 1.
Comparison of annotation of Arabidopsis thaliana (Araport11) representative peptide sequences
| Hayai-Annotation Plants a (4 best hits) | Hayai-Annotation Plants a (20 best hits) | Blast2GO b (Default) | TRAPID c | |
|---|---|---|---|---|
| TP | 103 927 | 105 283 | 105 470 | 59 071 |
| FP | 13 589 | 14 379 | 34 841 | 632 636 |
| FN | 24 557 | 23 201 | 23 014 | 69 413 |
| Specificity (%) | 88.4 | 88.0 | 75.2 | 8.5 |
| Sensitivity (%) | 80.9 | 81.9 | 82.1 | 46.0 |
| Accuracy (%) | 84.7 | 85.0 | 78.6 | 27.3d |
| P valuee | 0.6330 | 0.6168 | Reference | 6.2e-07 |
| Running time (Total) | 4 m 11 s | 11 m 58 s | 21 h 28 m* | 6 h 11 m |
| Running time (Alignment) | 3 m 57 s | 11 m 07 s | 17 h 35 m* | NS |
| Running time (Annotation) | 14 s | 51 s | 3 m 53 s | NS |
Note: UniProt-GOA was used as GO references for the purpose of accuracy calculation. Mercator4 only generated EC numbers. FunctionAnnotator, after 10 days, retrieved no results.
Hayai-Annotation Plants were run on Macintosh laptop (2.2 GHz, 2 cores, 8 GB RAM).
Blast2GO was set to cloud computing. *BLAST+ was performed on a server (2.40 GHz, 16 cores, 32 GB RAM) with query FASTA file split in 10 (parallel run) in order to increase speed. XML file were uploaded to Blast2GO.
TRAPID, transfer from best similarity hit where chosen to assign functional annotation. It is a web-based system; thus, speed varies accordingly to demand.
Significative different accuracy compared with Blast2GO, significant level 0.01.
P value: Chi-square test calculated based on the hypothesis that Hayai-Annotation Plants and Blast2GO have the same accuracy (Ho), against alternate hypothesis that Hayai-Annotation Plants and Blast2GO have a different accuracy (Ha).
TP: True Positive, FP: False Positive, FN: False Negative, NS: non-specified.