Table 6. Sensitivities and average scores of the different HMMs for evolutionarily distant retroviral LTRs at high specificity.
Thre-shold | Beta exo-genous | HML cons | Beta opossum | Alpha-beta chicken | Alpha exo-genous | Lenti | Delta | Spuma | Epsilon | Gamma exo-genous | HERV-gamma cons | Gamma opossum | |
Size of test set | 3 | 23 | 89 | 47 | 3 | 8 | 8 | 7 | 4 | 3 | 69 | 474 | |
Average length of LTRs | 382 | 728 | 332 | 290 | 325 | 412 | 700 | 1177 | 809 | 554 | 630 | 448 | |
Hml | 5 | −14 | 40* | −43 | −49 | −31 | −11 | −16 | −15 | −33 | −20 | −13 | −34 |
0 | 100% | 1.1% | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | ||
Gamma | 5 | −14 | 1.9 | −13 | −17 | −23 | −0.41 | −0.14 | −6.6 | −8.2 | 12* | 23* | −9.9 |
0 | 22% | 3.4% | 0 | 0 | 2/8 | 1/8 | 0 | 0 | 3/3 | 97% | 9.1% | ||
Beta | 7 | 2.3 | 18* | −5.8 | 1.5* | 16* | 3.7** | 8.6* | 1.8 | −11 | −8.1 | 2.1 | −6.4 |
1/3 | 100% | 9.0% | 32% | 3/3 | 1/8 | 4/8 | 1/7 | 0 | 0 | 12% | 1.9% | ||
Lenti | 3 | −15 | −5.2 | −18 | −17 | −11 | 34* | −7.1 | −14 | −18 | −7.8 | −6.5 | −16 |
0 | 8.7% | 0 | 0 | 0 | 100% | 0 | 0 | 0 | 0 | 4.3% | 3.0% | ||
General | 4 | −2.2 | 9.4* | −5.9 | −1.9* | 5.0* | 5.6** | 3.0* | −1.1 | −8.5 | −3.6* | 7.5* | −4.8 |
1/3 | 96% | 15% | 30% | 2/3 | 5/8 | 4/8 | 0 | 0 | 0 | 86% | 10% |
The thresholds were the same as in the RepeatMasker comparison in Table 2. (*) indicates that at least some LTRs in the LTR set are in the training set of the HMM model. (**) indicates that the set was used in the test set. The “beta”, “alphabeta”, “delta”, “spuma” and “gamma” categories contained ERVs and/or XRVs, further described in [1].