DeepCORE: An interpretable multi-view deep neural network model to detect co-operative regulatory elements.

Area of Science:

• Genomics
• Epigenetics
• Computational Biology

Background:

• Gene transcription regulation is complex, involving multiple interacting elements.
• Understanding these regulatory networks is crucial for deciphering biological traits and diseases.

Approach:

• Developed DeepCORE, a novel multi-view attention-based deep neural network.
• Modeled relationships between genetic, epigenetic, and transcriptional patterns.
• Applied DeepCORE to predict transcriptomes across 25 cell lines, outperforming existing algorithms.

Key Points:

• DeepCORE identifies co-operative regulatory elements (COREs) by integrating multi-modal data.
• Attention mechanisms provide interpretable insights into regulatory element locations and correlations.
• Discovered COREs are significantly enriched with known promoters and enhancers.

Conclusions:

• DeepCORE offers a powerful new approach for dissecting gene regulatory networks.
• Novel regulatory elements identified by DeepCORE exhibit characteristic epigenetic signatures.
• This method advances our understanding of gene regulation and its role in cellular function and disease.