Thresholds and effectiveness of scores derived from human–mouse–rat genome alignments for finding CRMs in the reference set for the HBB complex
Binary discrimination by position | Discrimination of intervalsb | |||||||
|---|---|---|---|---|---|---|---|---|
| Method | Threshold | Sn, Spa | Threshold | Snint | True discovery rate | |||
| MCS | 1.55 | 0.55 | 3.3 | 0.44 | 0.54 | |||
| PhastCons | 0.031 | 0.50 | 0.13 | 0.50 | 0.54 | |||
| Regulatory potential | -0.006 | 0.61 | 0 | 0.78 | 0.63 | |||
| Regulatory potential, no HBBc | -0.007 | 0.60 | 0 | 0.78 | 0.59 | |||
a Sn indicates sensitivity; Sp, specificity. Optimal Sn and Sp are determined at the crossover point and thus are equal.
b The intervals are evaluated by the maximum score within each interval.
c RP scores computed using a training set that excludes CRMs from the HBB complex.