Skip to main content

Table 1 Computational approaches to predicting gene targets of enhancer elements

From: Enhancer variants: evaluating functions in common disease

Reference or method Input data required Gene expression Linear model Number of genes with predictions (per cell line) FDR Species Publically available
Nearest gene None Not considered Nearest gene NA ~40% to 73% Any NA
Nearest expressed gene Gene expression Considered Nearest expressed gene NA ~53% to 77% Any NA
Ernst et al.[21] H3K4me1, H3K4me2, H3K27ac, RNA-seq Considered Distance based (125 kb) NA Not determined Human No
Thurman et al.[66] DNase I hypersensitivity Not considered Distance based (500 kb) NA Not determined Human No
Sheffield et al.[70] DNase I hypersensitivity and RNA-seq Considered 100 kb NA Not determined Human Predicted interactions: []
Shen et al.[67] H3K4me1, H3K27ac, RNA Pol II Not considered Topological domain based 5,000 to 8,000 Not determined Mouse No
Andersson et al.[16] CAGE Considered 500 kb NA Not determined Human No
PreSTIGE [7] H3K4me1 Considered Distance (100 kb) and CTCF based 3,000 to 5,000 ~13% to 23% Human Predicted interactions: Method application:
PreSTIGEouse [68] H3K4me1 Considered Distance based (100 kb) 3,000 to 5,000 Not determined Mouse Predicted interactions: Method application:
IM-PET [69] H3K4me1, H3K27ac, H3K4me3 and RNA-seq* Considered Distance (2 Mb) 7,000 to 10,000 ~1% Human Method application:
  1. *Input data utilized in publication, other input options exist. CAGE, cap analysis of gene expression; CTCF, CCCTC-binding factor (zinc finger protein demonstrated to function as an insulator protein); FDR, false discovery rate; Mb, megabases; NA, not applicable; RNA-seq, RNA sequencing.