Predicting active enhancers with DNA methylation and histone modification.

Area of Science:

• Molecular Biology
• Genomics
• Epigenetics

Background:

• Enhancers are critical for gene regulation, producing enhancer RNAs (eRNAs).
• Current eRNA detection methods like CAGE-seq suffer from noise due to eRNA instability.
• Limited research exists on CAGE-seq noise and eRNA prediction models.

Purpose of the Study:

• To reduce false positives in eRNA identification.
• To develop robust eRNA prediction models.
• To address the challenge of noise in eRNA detection.

Main Methods:

• Adjusting the statistical distribution of expression levels to minimize noise.
• Developing prediction models integrating gene expression, DNA methylation, and histone modification data.
• Validating model performance using intra-cell and cross-cell prediction.

Main Results:

• A novel method effectively reduced noise in eRNA detection.
• Developed eRNA prediction models achieved high accuracy (AUC ~0.95 intra-cell, ~0.9 cross-cell).
• The models demonstrated robustness across different cellular contexts.

Conclusions:

• The proposed method successfully attenuates noise from stochastic RNA production.
• The eRNA prediction model shows significant accuracy and robustness.
• This work provides a valuable tool for accurate eRNA detection and prediction.