Table 3. Comparison of the performances of PASTEC, REPCLASS, and TECLASS for classification to TE order level.
Dataset (sequence #) | Performance | PASTEC | PASTEC mapped to REPCLASS order (1) | PASTEC mapped to TECLASS order (2) | REPCLASS (3) | TECLASS (4) |
Repbase-atha (318) | ||||||
Well classified | 71.4 | 79.7 | 93.7 | 85.5 | 97.5 | |
Misclassified | 10.2 | 1.9 | 2.5 | 2.8 | 1.9 | |
Not classified | 18.4 | 18.4 | 3.8 | 11.6 | 0.6 | |
Repbase-diff (5546) | ||||||
Well classified | 51.3 | 59.1 | 49.7 | 66.9 | 47.7 | |
Misclassified | 10.9 | 3.1 | 3.5 | 11.1 | 50.3 | |
Not classified | 33.4 | 33.4 | 46.8 | 17.6 | 1.9 | |
Repbase-all (9665) | ||||||
Well classified | 22.4 | 31.9 | 8.6 | 12.4 | 40 | |
Misclassified | 15.8 | 6.2 | 6.7 | 33.3 | 59.7 | |
Not classified | 61.8 | 61.8 | 84.7 | 50.84 | 0.25 |
Note that the classification differs between the three tools. We therefore mapped the PASTEC classification results onto those for REPCLASS (1) and TECLASS (2). (2) Mapped onto TECLASS class I orders only. (3) order considered are: DNA transposon, LTR retrotransposon, helitron, non LTR retrotransposon. (4) order considered are only LTR, LINE/SINE.