Enhancer/gene relationships: Need for more reliable genome-wide reference sets

Area of Science:

• Genomics
• Molecular Biology
• Computational Biology

Background:

• Cellular functions are regulated by enhancers, cis-regulatory elements crucial for gene expression.
• Genetic variants in enhancers are linked to common diseases, highlighting the importance of understanding enhancer-gene (E/G) relationships.
• Identifying E/G links across cell types aids in deciphering genetic mechanisms of human diseases.

Purpose of the Study:

• To compare the performance of three recent methods (TargetFinder, Average-Rank, ABC model) for identifying enhancer-gene links.
• To evaluate these methods using established benchmark datasets (BENGI, CRiFF, CRiSPRi).
• To assess the current state of E/G relationship identification and identify needs for future research.

Main Methods:

• Comparative analysis of three computational methods: TargetFinder, Average-Rank, and the ABC model.
• Utilized three benchmark datasets: BENGI (combining 3D and eQTL data), CRiFF, and CRiSPRi (genetic screening data).
• Evaluated method performance based on accuracy and reliability across different reference sets.

Main Results:

• No single method consistently outperformed the others across all three benchmark references.
• CRiFF and CRiSPRi reference sets showed higher reliability but have limitations (not genome-wide, cell-type specific).
• The BENGI reference set is genome-wide but may contain a significant number of false positives.

Conclusions:

• Current functional link methods for identifying enhancer-gene relationships lack a definitive best performer.
• There is a critical need for new, reliable, and genome-wide reference datasets for evaluating E/G identification methods.
• Future efforts should focus on generating robust reference data rather than solely developing new identification algorithms.