EPIP: a novel approach for condition-specific enhancer-promoter interaction prediction

Area of Science:

• Genomics and Bioinformatics
• Molecular Biology
• Computational Biology

Background:

• Identifying enhancer-promoter interactions (EPIs) is crucial for understanding gene transcriptional regulation.
• Current experimental methods for EPI identification are costly.
• Existing computational tools often lack performance in predicting condition-specific EPIs.

Purpose of the Study:

• To develop a novel computational method for reliable prediction of EPIs, with a focus on condition-specific interactions.
• To provide a tool that performs well even with limited sample data.

Main Methods:

• Development of a novel computational method named EPIP.
• Validation of EPIP across multiple cell lines (more than eight).
• Performance evaluation using area under the receiver operating characteristic curve (AUC-ROC) and area under the precision-recall curve (AUC-PR).

Main Results:

• EPIP achieved an average AUC-ROC of 0.95 and an average AUC-PR of 0.73 across tested cell lines.
• EPIP demonstrated high accuracy in identifying condition-specific EPIs, correctly identifying 99.26% of them.
• EPIP outperformed two recently developed computational methods in accuracy.

Conclusions:

• EPIP is a reliable and accurate computational tool for predicting enhancer-promoter interactions, including condition-specific ones.
• The method is effective for samples with both limited and abundant data.
• EPIP is freely available for use, facilitating research in gene regulation.