A survey on computational methods for enhancer and enhancer target predictions

Enhancers are important cis-regulatory elements that play critical roles in a wide range of cellular processes by enhancing expression of target genes through promoter-enhancer loops. There are many interesting biological questions about enhancers, including their evolution and the relationships between their dysregulation and genetic diseases. The recent developments of experimental methods such as high-throughput reporter assays and ChIA-PET have enabled large-scale identification of enhancers and their targets. However, the current lists of identified enhancers and enhancer targets remain incomplete and unreliable due to the high noise level or low resolution of these methods. As a result, computational methods have emerged as an alternative for predicting the genomic locations of enhancers and their target genes. These methods have used a variety of features for predicting enhancers, including sequence motifs and epigenomic modifications. Potential enhancer targets have been predicted using activity correlations, distance constraints, and other features. Both prediction tasks are non-trivial due to cell-type specificity of enhancer activities, lack of definite orientation and distance of an enhancer from its target genes, insufficient known examples for training computational models, and other complexities. In this survey, we discuss the current computational methods for these two prediction tasks and analyze their pros and cons. We also point out obstacles of computational prediction of enhancers and enhancer targets in general, and suggest future research directions.