Table 2.

Overview of data and features used for enhancer identification

Data sources	Feature example	Advantage	Disadvantage	Representative methods
Evolutionary conservation	Conserved motifs across species	Easy to compute	Insufficient information for predicting enhancer's tissue-specific activity	[20]
Histone marks	ChIP-seq from H3K4me1	Provides cell-line-/tissue-specific information that characterize enhancers and also different categories of enhancers (e.g. poised versus active)	Different cell lines/tissues are associated with different combination of histone marks	[21, 28, 33, 34]
TFBSs	ChIP-seq from P300	Provides cell-line-/tissue-specific information that characterize enhancers. High-resolution data for testing activity of enhancer-related TFs	Not available for many cell lines/tissues	[23, 29]
Open chromatin	DHS	High discriminative capacity when combined with other data types, e.g. P300-binding sites	Regions with enriched DHS activity do not necessarily correspond to enhancers	[25]
Sequence characteristics	Kmers of size 5	Easy to compute	Insufficient information for predicting enhancers’ activity across different tissues	[39, 51]
eRNA expression	CAGE data	High accuracy	eRNA regulation mechanisms are unknown, and not all of the enhancers are known to produce eRNAs	[40]
Enhancer-screening data	STARR-seq	High accuracy for testing enhancer activity	Not useful for ab initio discovery of enhancers	[42, 43, 52]