Skip to main content

Table 1 Tissues for which the promoter models produce the most robust sets of predictions for enhancers

From: Sequence signatures extracted from proximal promoters can be used to predict distal enhancers

Tissue Number of enhancer predictions Fraction of loci with enhancer predictions Prediction scores AUC
  Fold enrichment P-value Fold enrichment P-value P-value  
Adrenal gland 2.70 1.39 × 10-73 1.13 4.92 × 10-2 2.64 × 10-5 0.83
Colorectal adenocarcinoma 4.63 1.79 × 10-155 1.25 2.12 × 10-3 4.11 × 10-8 0.85
Heart 7.78 1.21 × 10-277 1.31 2.82 × 10-4 2.19 × 10-13 0.93
Kidney 2.76 1.84 × 10-91 1.34 1.07 × 10-6 1.48 × 10-7 0.89
Liver 4.69 4.99 × 10-93 1.40 4.75 × 10-4 9.08 × 10-5 0.92
Lung 6.53 1.22 × 10-220 1.50 9.60 × 10-8 3.11 × 10-6 0.91
Placenta 2.99 4.18 × 10-83 1.28 4.13 × 10-4 5.76 × 10-4 0.83
Prefrontal cortex 1.23 2.27 × 10-10 1.21 3.47 × 10-6 2.71 × 10-2 0.66
Spinal cord 1.50 8.27 × 10-38 1.15 2.39 × 10-4 1.38 × 10-2 0.72
Tongue 1.92 3.17 × 10-62 1.10 3.85 × 10-2 4.82 × 10-5 0.78
  1. Only promoter-based models yielding AUC greater than 0.6 were considered in this analysis. The performance of each model in predicting enhancers was assessed by the significance of the difference between the relative number of enhancer predictions in the loci of highly expressed and lowly expressed genes with respect to the total number of scanned sequences, the significance of the difference between the fraction of loci of highly and lowly expressed genes comprising enhancer predictions, and the significance of the difference between the scores of enhancer predictions in predictions in the loci of highly and lowly expressed genes (see Materials and methods). P-values were computed using Fisher’s exact test and Wilcoxon rank-sum test.