Table 1.
Evaluation of SMARTest Accuracy
| Sequence | Experimentally defined S/MARs Position (kb) | SMARTest predictions Position (kb) | MAR-Finder predictions Position (kb) | |
|---|---|---|---|---|
| Description | Length (kb) | |||
| Oryza sativa putative ADP-glucose pyrophosphorylase | 30.034 | 0.0–1.2 | — | — |
| subunit SH2 and putative NADPH-dependent | 5.4–7.4 | 6.5–7.0 | — | |
| reductase A1 genes (U70541; Avramova et al. | 15.2–15.7 | 15.7–15.9 | ||
| 1998) | 16.2–16.6 | |||
| 17.3–18.5 | 17.6–18.3 | 17.5–18.4 | ||
| 20.0–23.1 | 19.6–20.1 | 19.8–20.4 | ||
| 20.7–21.3 | 21.3–21.5 | |||
| 23.6–23.9 | 23.9–24.2 | |||
| 25.0–25.4 | 24.7–25.1 | |||
| 27.5–27.9 | ||||
| Sorghum bicolor ADP-glucose pyrophosphorylase | 42.446 | 0.0–1.5 | — | — |
| subunit SH2, NADPH-dependent reductase A1-a | 7.1–9.7 | — | — | |
| and NADPH-dependent reductase A1-b genes. | 21.3–21.9 | |||
| (AF010283; Avramova et al. 1998) | 22.4–24.7 | 22.9–24.0 | 23.2–24.2 | |
| 27.3–27.6 | 26.9–27.5 | |||
| 32.5–33.7 | — | — | ||
| 41.6–42.3 | — | — | ||
| Sorghum bicolor BAC clone 110K5 (AF124045; | 78.195 | ∼0.9 | — | — |
| Tikhonov et al. 2000) | ∼5.8 | — | — | |
| ∼6.3 | — | — | ||
| ∼9.3 | — | — | ||
| ∼15.0 | 15.1–15.8 | — | ||
| ∼18.5 | — | — | ||
| ∼21.9 | 21.7–22.0 | — | ||
| ∼23.3 | — | — | ||
| ∼25.6 | — | — | ||
| ∼29.1 | — | — | ||
| ∼34.6 | — | — | ||
| ∼44.1 | 44.1–44.5 | — | ||
| ∼48.5 | 47.9–49.5 | 47.9–49.4 | ||
| ∼57.9 | — | — | ||
| ∼62.9 | 63.1–63.7 | — | ||
| ∼67.1 | — | — | ||
| ∼69.3 | — | — | ||
| ∼73.7 | 74.3–74.7 | — | ||
| Human alpha-1-antitrypsin and corticosteroid | 30.461 | 2.6–6.3 | 5.5–6.0 | 3.0–3.2 |
| binding globulin intergenic region (AF156545; | 5.1–6.0 | |||
| Rollini et al. 1999) | 22.0–30.4 | 25.7–26.2 | 24.9–25.3 | |
| 27.5–27.8 | 25.5–25.8 | |||
| 26.2–26.4 | ||||
| 27.5–28.2 | ||||
| Human protamine locus (U15422 & AC00247; | 53.060 | 8.8–9.7 | — | 8.0–8.9* |
| Kramer et al. 1998) | 32.6–33.6 | — | 33.9–34.8* | |
| 37.2–39.4 | — | 37.7–38.6* | ||
| 51.8–53.0 | — | —* | ||
| Human beta-globin locus (22754 & U01317; Jarman | 75.955 | 1.5–3.0 | — | — |
| and Higgs 1988) | 15.6–19.0 | 18.0–18.4 | 15.5–16.0 | |
| 18.0–18.4 | ||||
| 34.4–34.9 | ||||
| 44.7–52.7 | — | 50.6–50.8 | ||
| 56.6–57.1 | 56.5–57.2 | |||
| 60.0–70.0 | 59.8–60.3 | 58.1–58.5 | ||
| 65.6–66.0 | 63.0–63.6 | |||
| 67.6–67.9 | 68.7–69.3 | |||
| 68.8–69.1 | ||||
| Sum (kb) | 310.151 | >56.1 | 14.5 | 13.8 |
| Total numbers: | 37 | 28 | 25 | |
| True positives [number of experimentally defined S/MARs found] | — | 19 [14] | 20 [12] | |
| False positives | — | 9 | 5 | |
| False negatives | — | 23 | 25 | |
| Specificity | — | 19/28 = >68% | 20/25 = >80% | |
| Sensitivity | — | 14/37 = >38% | 12/37 = >32% | |
Six different genomic sequences, three plant and three human sequences, for which experimentally defined S/MARs are known were analyzed with SMARTest to evaluate the accuracy of in silico predictions. For comparison, the same sequences were analyzed with the MAR-Finder program. True positive matches are printed in bold, minus (−) indicates false negative matches. Some of the longer (up to 10 kb) experimentally defined S/MARs contained more than one in silico prediction, each of them was counted as true positive match. Therefore, the number of true in silico predictions is higher than the number of experimentally defined S/MARs found. Specificity is defined as the ratio of true positive predictions, whereas sensitivity is defined as the ratio of experimentally defined S/MARs found.
A T-rich rule excluded using MAR-Finder.