Table 3 Comparative testing results of POLYAR, polya_svm and polyadq rograms in [-300:+300] regions around the mapped PAS-strong, PAS-weak and PAS-less CS/poly (A) sites
From: POLYAR, a new computer program for prediction of poly(A) sites in human sequences
Data Set | Programs | TP | FN | FP | Predicted CSs, total | Sn,% * | Sp,% * |
|---|---|---|---|---|---|---|---|
PAS-strong (5225 sequences) | POLYAR | 4221 | 1004 | 2135 | 6356 | 80.8 | 66.4 |
| Â | polya_svm | 3431 | 1794 | 3206 | 6637 | 65.7 | 51.7 |
| Â | polyadq | 2804 | 2421 | 1177 | 3981 | 53.7 | 70.4 |
PAS-weak (2475 sequences) | POLYAR | 624 | 1851 | 1571 | 2195 | 25.2 | 28.4 |
| Â | polya_svm | 645 | 1830 | 1417 | 2185 | 26.1 | 29.5 |
PAS-less (561 sequences) | POLYAR | 76 | 485 | 544 | 518 | 13.5 | 14.7 |
| Â | polya_svm | 42 | 519 | 223 | 292 | 7.5 | 14.4 |