Table 3 Compared accuracy of the ERPIN and POLYADQ programs for the prediction of EMBL annotated poly(A) sites, and of EST-derived weak poly(A) sites identified in this study. The negative set for SP and CC calculation is "UTR shuffled" from Table 2.
From: Sequence determinants in human polyadenylation site selection
Program | Prediction of: | Sensitivity SN | Specificity SP | Accuracy CC |
|---|---|---|---|---|
Erpin | EMBL annotated sites | 55.91 % | 85.38 % | 0.494 |
Polyadq | Â | 55.70 % | 77.81 % | 0.415 |
Erpin | Weak sites | 31.28 % | 80.21 % | 0.262 |
Polyadq | Â | 30.54 % | 70.45 % | 0.171 |