Learning Cell-Type-Specific Gene Regulation Mechanisms by Multi-Attention Based Deep Learning With Regulatory Latent...

Area of Science:

• Genomics
• Computational Biology
• Epigenetics

Background:

• Epigenetic gene regulation is crucial for controlling gene expression.
• Current models often use single epigenetic markers, limiting their ability to capture complex regulatory mechanisms.
• Modeling transcriptional regulation with multiple epigenetic markers remains a challenge.

Purpose of the Study:

• To develop a novel deep learning model for characterizing complex gene regulation mechanisms using multiple epigenetic markers.
• To improve the accuracy of gene expression prediction by integrating multi-marker epigenetic data.
• To identify cell-type-specific gene expression control mechanisms.

Main Methods:

• Proposed a multi-attention based deep learning model.
• Integrated multiple epigenetic markers for gene regulation modeling.
• Utilized 18 cell line multi-omics data for experimental validation.

Main Results:

• The proposed model achieved higher accuracy in predicting gene expression levels compared to state-of-the-art methods.
• Successfully identified cell-type-specific gene expression control mechanisms.
• Identified genes enriched for specific cell types based on function and epigenetic regulation.

Conclusions:

• Multi-marker epigenetic data integration via deep learning enhances gene expression prediction accuracy.
• The model provides insights into complex, cell-type-specific gene regulatory networks.
• This approach facilitates the discovery of functionally relevant genes in specific cell types.